AM62A3: Models compiled with edgeai-tidl-tools cause segmentation fault on AM62a

Stefan Werner

Part Number: AM62A3

Tool/software:

Hello, I am using a AM62a target device where offloading models to the accelerator causes a segmentation fault.

- I have validated the "out-of-box" examples within a docker container that I use to compile the artifacts

- I have compiled the osrt_python/tfl and osrt_python/ort examples successfully within the container (these are the files that I am transfering to the target)

- I have tested the inference in the docker container without offloading

- I have tested the inference without offloading on my target device

Only when I enable offloading on the AM62a device, I get the segmentation fault. This is the case for the osrt_python/tfl as well as the osrt_python/ort models.

- I am using the release tag 10_00_07_00 within the docker container as well as on the target device

- I have recently updated the target device's TIDL version following this explanation edgeai-tidl-tools/docs/backward_compatibility.md at 10_00_07_00 · TexasInstruments/edgeai-tidl-tools

On a side note: is the description in edgeai-tidl-tools/docs/backward_compatibility.md at 10_00_07_00 · TexasInstruments/edgeai-tidl-tools sufficient to upgrade / install the edgeai-tidl-tool on the target device? I've read something about an RTOS SDK version at some point. Is this also something I need to upgrade? If so, how?

What can I do about this? How do I approach this problem? Thanks for the help!

1 month ago

0 Stefan Werner 1 month ago

Prodigy 30 points

Output of python tflrt_delegate.py:

root@am62dl:/opt/edgeai-tidl-tools/examples/osrt_python/tfl# python3 tflrt_delegate.py
Running 4 Models - ['cl-tfl-mobilenet_v1_1.0_224', 'ss-tfl-deeplabv3_mnv2_ade20k_float', 'od-tfl-ssd_mobilenet_v2_300_float', 'od-tfl-ssdlite_mobiledet_dsp_320x320_coco']

Running_Model : cl-tfl-mobilenet_v1_1.0_224

Number of subgraphs:1 , 34 nodes delegated out of 34 nodes

Segmentation fault (core dumped)

0 Reese Grimsley 1 month ago in reply to Stefan Werner

TI__Genius 10636 points

Hello Stefan,

We will figure out where this is coming from. My first suspicion is related to tool versions

Stefan Werner said:
- I am using the release tag 10_00_07_00 within the docker container as well as on the target device

TIDL Tools versions with odd numbers are designated as ones portable to the previous SDK, in this case 9.2. Is that the version of the SDK that you have on your AM62A installation?

You should have $EDGEAI_SDK_VERSION set to a string similar to 09_02 or 9.2 defined within your linux environment (defined by an auto-run script on login)
You should have updated the firmware, OSRT components (e.g. TFLite and ONNXRT libs), and other ti libraries like libtivision_apps.so using the steps mentioned in the backwards_compatibility.md doc
- it sounds like you have done this, but simply verifying here
- Ensure the $SOC environment variable was set to 'am62a'. This should have also been handled by the auto-run script on login

If you are seeing seg faults on any model, then my estimation is that something was not correctly updated. Your approach on testing different components is isolating this to the TIDL stack, so it is helpful.

I am also curious why your device's hostname is root@am62dl, but perhaps that was intentional and we can ignore.

Suggested steps for collecting more info/logs:

pass debug_level=2 to the runtime when creating the model. It should be sufficient to set this in examples/osrt_python/common_utils.py
On target, run /opt/vx_app_arm_remote_log.out in the background before starting your script
Run the python application from gdb, and check the backtrace for the thread that seg-faulted
Run a `pip3 freeze | grep -i "tflite\|onnx\|tidl" and share the package versions

Stefan Werner said:
On a side note: is the description in edgeai-tidl-tools/docs/backward_compatibility.md at 10_00_07_00 · TexasInstruments/edgeai-tidl-tools sufficient to upgrade / install the edgeai-tidl-tool on the target device? I've read something about an RTOS SDK version at some point. Is this also something I need to upgrade? If so, how?

Yes, the instructions here are sufficient to upgrade the TIDL stack (not just edgeai-tidl-tools) on the previous SDK with latest bugfixes and changes, with one caveat -- the memory map between the EVM and your hardware platform must be compatible. If you are on the starter kit EVM, ignore this point.

I do not think the RTOS SDK (probably PSDK RTOS) is necessary here, but please point me towards this note if you happen across it again. If you needed to change the memory map for your custom hardware, this would be relevant. Note that for AM62A, we have a 'firmware-builder' tool that occupies same function as PSDK RTOS SDK.

BR,
Reese

0 Stefan Werner 1 month ago in reply to Reese Grimsley

Prodigy 30 points

Hello Reese,

thanks for the help, appreciated!

Reese Grimsley said:
TIDL Tools versions with odd numbers are designated as ones portable to the previous SDK, in this case 9.2. Is that the version of the SDK that you have on your AM62A installation?

The EDGEAI_SDK_VERSION is set to 09_00_00. Since I've tried to update the target device to 10_00_07_00, I guess this is wrong, no? I've checked the setup_target_device.sh script that we've used to update the device and could not find anything related to updating this environment variable. Am I missing some steps here to properly update the device to be compatible with the models compiled with edgeai-tidl-tools version 10_00_07_00? On the target device we have used the 10_00_07_00 tag of the edgai-tidl-tools: "root@am62dl:/opt/edgeai-tidl-tools# git status HEAD detached at 10_00_07_00" .

Reese Grimsley said:
Ensure the $SOC environment variable was set to 'am62a'. This should have also been handled by the auto-run script on logi

The SOC variable is indeed set to am62a upon logging into the target device. This is also what we set to compile the model artifacts in the docker container.

Reese Grimsley said:
I am also curious why your device's hostname is root@am62dl, but perhaps that was intentional and we can ignore.

Yes, this is just us renaming our device. This should not matter at all.

Reese Grimsley said:
Run the python application from gdb, and check the backtrace for the thread that seg-faulted

This is the output of GDB when using the option *thread apply all bt*. Does anything suspicious come to mind here?

(gdb) run tflrt_delegate.py 
Starting program: /usr/bin/python3 tflrt_delegate.py
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/libthread_db.so.1".
warning: Cannot parse .gnu_debugdata section; LZMA support was disabled at compile time
[New Thread 0xfffff57cf120 (LWP 81481)]
[New Thread 0xfffff2fbf120 (LWP 81482)]
[New Thread 0xfffff07af120 (LWP 81483)]
Running 4 Models - ['cl-tfl-mobilenet_v1_1.0_224', 'ss-tfl-deeplabv3_mnv2_ade20k_float', 'od-tfl-ssd_mobilenet_v2_300_float', 'od-tfl-ssdlite_mobiledet_dsp_320x320_coco']


Running_Model :  cl-tfl-mobilenet_v1_1.0_224

 Number of subgraphs:1 , 34 nodes delegated out of 34 nodes 
 

Thread 1 "python3" received signal SIGSEGV, Segmentation fault.
0x0000000500000004 in ?? ()
(gdb) thread apply all bt

Thread 4 (Thread 0xfffff07af120 (LWP 81483) "python3"):
#0  0x0000fffff7aec96c in ?? () from /lib/libc.so.6
#1  0x0000fffff7aef698 in pthread_cond_wait () from /lib/libc.so.6
#2  0x0000fffff5c47488 in blas_thread_server () from /usr/lib/python3.10/site-packages/numpy/core/../../numpy.libs/libopenblas64_p-r0-9c1f2efe.3.20.so
#3  0x0000fffff7af02e8 in ?? () from /lib/libc.so.6
#4  0x0000fffff7b57c1c in ?? () from /lib/libc.so.6

Thread 3 (Thread 0xfffff2fbf120 (LWP 81482) "python3"):
#0  0x0000fffff7aec96c in ?? () from /lib/libc.so.6
#1  0x0000fffff7aef698 in pthread_cond_wait () from /lib/libc.so.6
#2  0x0000fffff5c47488 in blas_thread_server () from /usr/lib/python3.10/site-packages/numpy/core/../../numpy.libs/libopenblas64_p-r0-9c1f2efe.3.20.so
#3  0x0000fffff7af02e8 in ?? () from /lib/libc.so.6
#4  0x0000fffff7b57c1c in ?? () from /lib/libc.so.6

Thread 2 (Thread 0xfffff57cf120 (LWP 81481) "python3"):
#0  0x0000fffff7aec96c in ?? () from /lib/libc.so.6
#1  0x0000fffff7aef698 in pthread_cond_wait () from /lib/libc.so.6
#2  0x0000fffff5c47488 in blas_thread_server () from /usr/lib/python3.10/site-packages/numpy/core/../../numpy.libs/libopenblas64_p-r0-9c1f2efe.3.20.so
#3  0x0000fffff7af02e8 in ?? () from /lib/libc.so.6
#4  0x0000fffff7b57c1c in ?? () from /lib/libc.so.6

Thread 1 (Thread 0xfffff7ff5020 (LWP 81477) "python3"):
#0  0x0000000500000004 in ?? ()
#1  0x0000ffffecc14b24 in tflite::Subgraph::AddNodeWithParameters(std::vector<int, std::allocator<int> > const&, std::vector<int, std::allocator<int> > const&, std::vector<int, std::allocator<int> > const&, char const*, unsigned long, void*, TfLiteRegistration const*, int*) () from /usr/lib/python3.10/site-packages/tflite_runtime/_pywrap_tensorflow_interpreter_wrapper.so
#2  0x0000ffffecc17718 in tflite::Subgraph::ReplaceNodeSubsetsWithDelegateKernels(TfLiteRegistration, TfLiteIntArray const*, TfLiteDelegate*) () from /usr/lib/python3.10/site-packages/tflite_runtime/_pywrap_tensorflow_interpreter_wrapper.so
#3  0x0000ffffecc179e0 in tflite::Subgraph::ReplaceNodeSubsetsWithDelegateKernels(TfLiteContext*, TfLiteRegistration, TfLiteIntArray const*, TfLiteDelegate*) () from /usr/lib/python3.10/site-packages/tflite_runtime/_pywrap_tensorflow_interpreter_wrapper.so
#4  0x0000ffffec102ddc in tflite::tfl_delegate::DelegatePrepareInfer(TfLiteContext*, TfLiteDelegate*) () from /usr/lib/libtidl_tfl_delegate.so
--Type <RET> for more, q to quit, c to continue without paging--
#5  0x0000ffffecc18388 in tflite::Subgraph::ModifyGraphWithDelegateImpl(TfLiteDelegate*) () from /usr/lib/python3.10/site-packages/tflite_runtime/_pywrap_tensorflow_interpreter_wrapper.so
#6  0x0000ffffecc186cc in tflite::Subgraph::ModifyGraphWithDelegate(TfLiteDelegate*) () from /usr/lib/python3.10/site-packages/tflite_runtime/_pywrap_tensorflow_interpreter_wrapper.so
#7  0x0000ffffecbff6cc in tflite::impl::Interpreter::ModifyGraphWithDelegateImpl(TfLiteDelegate*) () from /usr/lib/python3.10/site-packages/tflite_runtime/_pywrap_tensorflow_interpreter_wrapper.so
#8  0x0000ffffecbdfdf0 in tflite::interpreter_wrapper::InterpreterWrapper::ModifyGraphWithDelegate(TfLiteDelegate*) () from /usr/lib/python3.10/site-packages/tflite_runtime/_pywrap_tensorflow_interpreter_wrapper.so
#9  0x0000ffffecbe7cf4 in pybind11::cpp_function::initialize<pybind11_init__pywrap_tensorflow_interpreter_wrapper(pybind11::module_&)::{lambda(tflite::interpreter_wrapper::InterpreterWrapper&, unsigned long)#24}, pybind11::object, tflite::interpreter_wrapper::InterpreterWrapper&, unsigned long, pybind11::name, pybind11::is_method, pybind11::sibling, char [60]>(pybind11_init__pywrap_tensorflow_interpreter_wrapper(pybind11::module_&)::{lambda(tflite::interpreter_wrapper::InterpreterWrapper&, unsigned long)#24}&&, pybind11::object (*)(tflite::interpreter_wrapper::InterpreterWrapper&, unsigned long), pybind11::name const&, pybind11::is_method const&, pybind11::sibling const&, char const (&) [60])::{lambda(pybind11::detail::function_call&)#3}::_FUN(pybind11::detail::function_call&) () from /usr/lib/python3.10/site-packages/tflite_runtime/_pywrap_tensorflow_interpreter_wrapper.so
#10 0x0000ffffecbf5c50 in pybind11::cpp_function::dispatcher(_object*, _object*, _object*) () from /usr/lib/python3.10/site-packages/tflite_runtime/_pywrap_tensorflow_interpreter_wrapper.so
#11 0x0000fffff7d144fc in ?? () from /usr/lib/libpython3.10.so.1.0
#12 0x0000fffff7cd50cc in _PyObject_MakeTpCall () from /usr/lib/libpython3.10.so.1.0
#13 0x0000fffff7cd7e94 in ?? () from /usr/lib/libpython3.10.so.1.0
#14 0x0000fffff7c8a9d0 in _PyEval_EvalFrameDefault () from /usr/lib/libpython3.10.so.1.0
#15 0x0000fffff7d9eae8 in ?? () from /usr/lib/libpython3.10.so.1.0
#16 0x0000fffff7cd52e4 in _PyObject_FastCallDictTstate () from /usr/lib/libpython3.10.so.1.0
#17 0x0000fffff7cd5504 in _PyObject_Call_Prepend () from /usr/lib/libpython3.10.so.1.0
#18 0x0000fffff7d36430 in ?? () from /usr/lib/libpython3.10.so.1.0
#19 0x0000fffff7d2f7d0 in ?? () from /usr/lib/libpython3.10.so.1.0
#20 0x0000fffff7cd50cc in _PyObject_MakeTpCall () from /usr/lib/libpython3.10.so.1.0
#21 0x0000fffff7c8a6f8 in _PyEval_EvalFrameDefault () from /usr/lib/libpython3.10.so.1.0
#22 0x0000fffff7d9eae8 in ?? () from /usr/lib/libpython3.10.so.1.0
#23 0x0000fffff7c84d80 in _PyEval_EvalFrameDefault () from /usr/lib/libpython3.10.so.1.0
#24 0x0000fffff7d9eae8 in ?? () from /usr/lib/libpython3.10.so.1.0
#25 0x0000fffff7d9eca4 in PyEval_EvalCode () from /usr/lib/libpython3.10.so.1.0
#26 0x0000fffff7ddb810 in ?? () from /usr/lib/libpython3.10.so.1.0
#27 0x0000fffff7ddba38 in ?? () from /usr/lib/libpython3.10.so.1.0
#28 0x0000fffff7ddbb58 in ?? () from /usr/lib/libpython3.10.so.1.0
#29 0x0000fffff7ddd848 in _PyRun_SimpleFileObject () from /usr/lib/libpython3.10.so.1.0
#30 0x0000fffff7dddd1c in _PyRun_AnyFileObject () from /usr/lib/libpython3.10.so.1.0
--Type <RET> for more, q to quit, c to continue without paging--
#31 0x0000fffff7df9850 in Py_RunMain () from /usr/lib/libpython3.10.so.1.0
#32 0x0000fffff7df9f9c in Py_BytesMain () from /usr/lib/libpython3.10.so.1.0
#33 0x0000fffff7a9b230 in ?? () from /lib/libc.so.6
#34 0x0000fffff7a9b30c in __libc_start_main () from /lib/libc.so.6
#35 0x0000000000400870 in _start ()

Reese Grimsley said:
Run a `pip3 freeze | grep -i "tflite\|onnx\|tidl" and share the package versions

root@am62dl:/opt/edgeai-tidl-tools/examples/osrt_python/tfl# pip3 freeze | grep -i "tflite\|onnx\|tidl"
onnxruntime-tidl @ file:///home/root/arago_j7_pywhl/onnxruntime_tidl-1.14.0%2B10000005-cp310-cp310-linux_aarch64.whl
tflite-runtime @ file:///home/root/arago_j7_pywhl/tflite_runtime-2.12.0-cp310-cp310-linux_aarch64.whl

Reese Grimsley said:
On target, run /opt/vx_app_arm_remote_log.out in the background before starting your script

I am not sure if I have done this correctly, but here goes the output after running the scripts once or twice.

[C7x_1 ] 2322568.032723 s: UDMA: Init ... Done !!!
[C7x_1 ] 2322568.032735 s: MEM: Init ... !!!
[C7x_1 ] 2322568.032747 s: MEM: Created heap (DDR_LOCAL_MEM, id=0, flags=0x00000004) @ b2000000 of size 117440512 bytes !!!
[C7x_1 ] 2322568.032776 s: MEM: Init ... Done !!!
[C7x_1 ] 2322568.032788 s: IPC: Init ... !!!
[C7x_1 ] 2322568.032800 s: IPC: 3 CPUs participating in IPC !!!
[C7x_1 ] 2322568.033017 s: IPC: Waiting for HLOS to be ready ... !!!
[C7x_1 ] 2322568.054528 s: IPC: HLOS is ready !!!
[C7x_1 ] 2322568.054614 s: IPC: Init ... Done !!!
[C7x_1 ] 2322568.054629 s: APP: Syncing with 2 CPUs ..

Reese Grimsley said:
pass debug_level=2 to the runtime when creating the model. It should be sufficient to set this in examples/osrt_python/common_utils.py

I've compiled the models in our container again and set the logging level. This is the captured output.

logging.log

Another question for the setup_target_device.sh script. I dont quite understand the instructions for the TISDK_IMAGE environment variable. How can I tell whether I need to set adas or edgeai here? What is the difference between EVM boards and SK boards?

export TISDK_IMAGE=*adas or edgeai* // [adas for evm boards, edgeai for sk boards]

Also, do I need to update the C7x firmware as well? I've used TISDK_IMAGE=edgeai and not updated the C7x firmware so far.

export UPDATE_FIRMWARE_AND_LIB=1

Really appreciate the help. Is there any other information you need? Do you know if the tidl installation on our target device is broken / has the wrong version? What are the next steps?

Best Regards

0 Stefan Werner 1 month ago in reply to Stefan Werner

Prodigy 30 points

Hello Reese,

I have run the setup_target_device.sh script again with the below environment variable to update the C7x firmware.

export UPDATE_FIRMWARE_AND_LIB=1

The osrt_python/tfl example no longer gives a segmentation fault (which is great!), but gets stuck when inferencing the model. This completely freezes the shell.

root@am62dl:/opt/edgeai-tidl-tools/examples/osrt_python/tfl# python3 tflrt_delegate.py 
Running 4 Models - ['cl-tfl-mobilenet_v1_1.0_224', 'ss-tfl-deeplabv3_mnv2_ade20k_float', 'od-tfl-ssd_mobilenet_v2_300_float', 'od-tfl-ssdlite_mobiledet_dsp_320x320_coco']


Running_Model :  cl-tfl-mobilenet_v1_1.0_224

 ****** In DelegatePrepare ****** 

 Number of subgraphs:1 , 34 nodes delegated out of 34 nodes 
 

 ****** In tidlDelegate::Init ****** 
************ in TIDL_subgraphRtCreate ************ 
 APP: Init ... !!!
MEM: Init ... !!!
MEM: Initialized DMA HEAP (fd=6) !!!
MEM: Init ... Done !!!
IPC: Init ... !!!
IPC: Init ... Done !!!
REMOTE_SERVICE: Init ... !!!
REMOTE_SERVICE: Init ... Done !!!
6214240.148224 s: GTC Frequency = 200 MHz
APP: Init ... Done !!!
6214240.150091 s:  VX_ZONE_INIT:Enabled
6214240.150135 s:  VX_ZONE_ERROR:Enabled
6214240.150150 s:  VX_ZONE_WARNING:Enabled
6214240.151546 s:  VX_ZONE_INIT:[tivxPlatformCreateTargetId:116] Added target MPU-0 
6214240.152033 s:  VX_ZONE_INIT:[tivxPlatformCreateTargetId:116] Added target MPU-1 
6214240.152484 s:  VX_ZONE_INIT:[tivxPlatformCreateTargetId:116] Added target MPU-2 
6214240.153477 s:  VX_ZONE_INIT:[tivxPlatformCreateTargetId:116] Added target MPU-3 
6214240.155439 s:  VX_ZONE_INIT:[tivxInitLocal:136] Initialization Done !!!
6214240.156932 s:  VX_ZONE_INIT:[tivxHostInitLocal:101] Initialization Done for HOST !!!

I captured the backtraces using gdb again:

Type "apropos word" to search for commands related to "word"...
Reading symbols from python3...
(No debugging symbols found in python3)
(gdb) run tflrt_delegate.py 
Starting program: /usr/bin/python3 tflrt_delegate.py
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/libthread_db.so.1".
warning: Cannot parse .gnu_debugdata section; LZMA support was disabled at compile time
[New Thread 0xfffff57cf120 (LWP 85454)]
[New Thread 0xfffff2fbf120 (LWP 85455)]
[New Thread 0xfffff07af120 (LWP 85456)]
Running 4 Models - ['cl-tfl-mobilenet_v1_1.0_224', 'ss-tfl-deeplabv3_mnv2_ade20k_float', 'od-tfl-ssd_mobilenet_v2_300_float', 'od-tfl-ssdlite_mobiledet_dsp_320x320_coco']


Running_Model :  cl-tfl-mobilenet_v1_1.0_224

 ****** In DelegatePrepare ****** 

 Number of subgraphs:1 , 34 nodes delegated out of 34 nodes 
 

 ****** In tidlDelegate::Init ****** 
************ in TIDL_subgraphRtCreate ************ 
 APP: Init ... !!!
MEM: Init ... !!!
MEM: Initialized DMA HEAP (fd=6) !!!
MEM: Init ... Done !!!
IPC: Init ... !!!
[New Thread 0xffffe73bf120 (LWP 85487)]
IPC: Init ... Done !!!
REMOTE_SERVICE: Init ... !!!
REMOTE_SERVICE: Init ... Done !!!
6214570.008009 s: GTC Frequency = 200 MHz
APP: Init ... Done !!!
6214570.008385 s:  VX_ZONE_INIT:Enabled
6214570.008457 s:  VX_ZONE_ERROR:Enabled
6214570.008506 s:  VX_ZONE_WARNING:Enabled
[New Thread 0xffffe6baf120 (LWP 85490)]
6214570.010860 s:  VX_ZONE_INIT:[tivxPlatformCreateTargetId:116] Added target MPU-0 
[New Thread 0xffffe639f120 (LWP 85491)]
6214570.012226 s:  VX_ZONE_INIT:[tivxPlatformCreateTargetId:116] Added target MPU-1 
[New Thread 0xffffe5b8f120 (LWP 85492)]
6214570.013536 s:  VX_ZONE_INIT:[tivxPlatformCreateTargetId:116] Added target MPU-2 
[New Thread 0xffffe537f120 (LWP 85493)]
6214570.014785 s:  VX_ZONE_INIT:[tivxPlatformCreateTargetId:116] Added target MPU-3 
6214570.014832 s:  VX_ZONE_INIT:[tivxInitLocal:136] Initialization Done !!!
6214570.015683 s:  VX_ZONE_INIT:[tivxHostInitLocal:101] Initialization Done for HOST !!!
^C
Thread 1 "python3" received signal SIGINT, Interrupt.
0x0000fffff7aec96c in ?? () from /lib/libc.so.6
(gdb) thread apply all bt

Thread 9 (Thread 0xffffe537f120 (LWP 85493) "python3"):
#0  0x0000fffff7aec96c in ?? () from /lib/libc.so.6
#1  0x0000fffff7aef698 in pthread_cond_wait () from /lib/libc.so.6
#2  0x0000ffffeaad2e98 in tivxQueueGet () from /usr/lib/libtivision_apps.so.9.2.0
#3  0x0000ffffeaacbff0 in ownTargetTaskMain () from /usr/lib/libtivision_apps.so.9.2.0
#4  0x0000ffffeaad2fbc in tivxTaskMain () from /usr/lib/libtivision_apps.so.9.2.0
#5  0x0000fffff7af02e8 in ?? () from /lib/libc.so.6
#6  0x0000fffff7b57c1c in ?? () from /lib/libc.so.6

Thread 8 (Thread 0xffffe5b8f120 (LWP 85492) "python3"):
#0  0x0000fffff7aec96c in ?? () from /lib/libc.so.6
#1  0x0000fffff7aef698 in pthread_cond_wait () from /lib/libc.so.6
#2  0x0000ffffeaad2e98 in tivxQueueGet () from /usr/lib/libtivision_apps.so.9.2.0
#3  0x0000ffffeaacbff0 in ownTargetTaskMain () from /usr/lib/libtivision_apps.so.9.2.0
#4  0x0000ffffeaad2fbc in tivxTaskMain () from /usr/lib/libtivision_apps.so.9.2.0
#5  0x0000fffff7af02e8 in ?? () from /lib/libc.so.6
#6  0x0000fffff7b57c1c in ?? () from /lib/libc.so.6

Thread 7 (Thread 0xffffe639f120 (LWP 85491) "python3"):
#0  0x0000fffff7aec96c in ?? () from /lib/libc.so.6
#1  0x0000fffff7aef698 in pthread_cond_wait () from /lib/libc.so.6
#2  0x0000ffffeaad2e98 in tivxQueueGet () from /usr/lib/libtivision_apps.so.9.2.0
#3  0x0000ffffeaacbff0 in ownTargetTaskMain () from /usr/lib/libtivision_apps.so.9.2.0
#4  0x0000ffffeaad2fbc in tivxTaskMain () from /usr/lib/libtivision_apps.so.9.2.0
#5  0x0000fffff7af02e8 in ?? () from /lib/libc.so.6
#6  0x0000fffff7b57c1c in ?? () from /lib/libc.so.6

Thread 6 (Thread 0xffffe6baf120 (LWP 85490) "python3"):
--Type <RET> for more, q to quit, c to continue without paging--
#0  0x0000fffff7aec96c in ?? () from /lib/libc.so.6
#1  0x0000fffff7aef698 in pthread_cond_wait () from /lib/libc.so.6
#2  0x0000ffffeaad2e98 in tivxQueueGet () from /usr/lib/libtivision_apps.so.9.2.0
#3  0x0000ffffeaacbff0 in ownTargetTaskMain () from /usr/lib/libtivision_apps.so.9.2.0
#4  0x0000ffffeaad2fbc in tivxTaskMain () from /usr/lib/libtivision_apps.so.9.2.0
#5  0x0000fffff7af02e8 in ?? () from /lib/libc.so.6
#6  0x0000fffff7b57c1c in ?? () from /lib/libc.so.6

Thread 5 (Thread 0xffffe73bf120 (LWP 85487) "python3"):
#0  0x0000fffff7b50b70 in select () from /lib/libc.so.6
#1  0x0000ffffeaafae10 in appIpcRpmsgRxTaskMain () from /usr/lib/libtivision_apps.so.9.2.0
#2  0x0000fffff7af02e8 in ?? () from /lib/libc.so.6
#3  0x0000fffff7b57c1c in ?? () from /lib/libc.so.6

Thread 4 (Thread 0xfffff07af120 (LWP 85456) "python3"):
#0  0x0000fffff7aec96c in ?? () from /lib/libc.so.6
#1  0x0000fffff7aef698 in pthread_cond_wait () from /lib/libc.so.6
#2  0x0000fffff5c47488 in blas_thread_server () from /usr/lib/python3.10/site-packages/numpy/core/../../numpy.libs/libopenblas64_p-r0-9c1f2efe.3.20.so
#3  0x0000fffff7af02e8 in ?? () from /lib/libc.so.6
#4  0x0000fffff7b57c1c in ?? () from /lib/libc.so.6

Thread 3 (Thread 0xfffff2fbf120 (LWP 85455) "python3"):
#0  0x0000fffff7aec96c in ?? () from /lib/libc.so.6
#1  0x0000fffff7aef698 in pthread_cond_wait () from /lib/libc.so.6
#2  0x0000fffff5c47488 in blas_thread_server () from /usr/lib/python3.10/site-packages/numpy/core/../../numpy.libs/libopenblas64_p-r0-9c1f2efe.3.20.so
#3  0x0000fffff7af02e8 in ?? () from /lib/libc.so.6
#4  0x0000fffff7b57c1c in ?? () from /lib/libc.so.6

Thread 2 (Thread 0xfffff57cf120 (LWP 85454) "python3"):
--Type <RET> for more, q to quit, c to continue without paging--
#0  0x0000fffff7aec96c in ?? () from /lib/libc.so.6
#1  0x0000fffff7aef698 in pthread_cond_wait () from /lib/libc.so.6
#2  0x0000fffff5c47488 in blas_thread_server () from /usr/lib/python3.10/site-packages/numpy/core/../../numpy.libs/libopenblas64_p-r0-9c1f2efe.3.20.so
#3  0x0000fffff7af02e8 in ?? () from /lib/libc.so.6
#4  0x0000fffff7b57c1c in ?? () from /lib/libc.so.6


Thread 1 (Thread 0xfffff7ff5020 (LWP 85440) "python3"):
#0  0x0000fffff7aec96c in ?? () from /lib/libc.so.6
#1  0x0000fffff7aef698 in pthread_cond_wait () from /lib/libc.so.6
#2  0x0000ffffeaad1fb4 in tivxEventWait () from /usr/lib/libtivision_apps.so.9.2.0
#3  0x0000ffffeaa94944 in ownContextSendCmd () from /usr/lib/libtivision_apps.so.9.2.0
#4  0x0000ffffeaab8b64 in ownNodeKernelInit () from /usr/lib/libtivision_apps.so.9.2.0
#5  0x0000ffffeaaa5fdc in vxVerifyGraph () from /usr/lib/libtivision_apps.so.9.2.0
#6  0x0000ffffebdccb20 in TIDLRT_create () from /usr/lib/libvx_tidl_rt.so
#7  0x0000ffffec103d74 in TIDL_subgraphRtCreate () from /usr/lib/libtidl_tfl_delegate.so
#8  0x0000ffffec102830 in tflite::tfl_delegate::tidlDelegate::Init(TfLiteContext*, TfLiteDelegateParams const*) () from /usr/lib/libtidl_tfl_delegate.so
#9  0x0000ffffec102980 in tflite::tfl_delegate::GetTIDLNodeRegistration()::{lambda(TfLiteContext*, char const*, unsigned long)#2}::_FUN(TfLiteContext*, char const*, unsigned long) () from /usr/lib/libtidl_tfl_delegate.so
#10 0x0000ffffecc14b24 in tflite::Subgraph::AddNodeWithParameters(std::vector<int, std::allocator<int> > const&, std::vector<int, std::allocator<int> > const&, std::vector<int, std::allocator<int> > const&, char const*, unsigned long, void*, TfLiteRegistration const*, int*) () from /usr/lib/python3.10/site-packages/tflite_runtime/_pywrap_tensorflow_interpreter_wrapper.so
#11 0x0000ffffecc17718 in tflite::Subgraph::ReplaceNodeSubsetsWithDelegateKernels(TfLiteRegistration, TfLiteIntArray const*, TfLiteDelegate*) () from /usr/lib/python3.10/site-packages/tflite_runtime/_pywrap_tensorflow_interpreter_wrapper.so
#12 0x0000ffffecc179e0 in tflite::Subgraph::ReplaceNodeSubsetsWithDelegateKernels(TfLiteContext*, TfLiteRegistration, TfLiteIntArray const*, TfLiteDelegate*) () from /usr/lib/python3.10/site-packages/tflite_runtime/_pywrap_tensorflow_interpreter_wrapper.so
#13 0x0000ffffec102e18 in tflite::tfl_delegate::DelegatePrepareInfer(TfLiteContext*, TfLiteDelegate*) () from /usr/lib/libtidl_tfl_delegate.so
#14 0x0000ffffecc18388 in tflite::Subgraph::ModifyGraphWithDelegateImpl(TfLiteDelegate*) () from /usr/lib/python3.10/site-packages/tflite_runtime/_pywrap_tensorflow_interpreter_wrapper.so
#15 0x0000ffffecc186cc in tflite::Subgraph::ModifyGraphWithDelegate(TfLiteDelegate*) () from /usr/lib/python3.10/site-packages/tflite_runtime/_pywrap_tensorflow_interpreter_wrapper.so
#16 0x0000ffffecbff6cc in tflite::impl::Interpreter::ModifyGraphWithDelegateImpl(TfLiteDelegate*) () from /usr/lib/python3.10/site-packages/tflite_runtime/_pywrap_tensorflow_interpreter_wrapper.so
#17 0x0000ffffecbdfdf0 in tflite::interpreter_wrapper::InterpreterWrapper::ModifyGraphWithDelegate(TfLiteDelegate*) () from /usr/lib/python3.10/site-packages/tflite_runtime/_pywrap_tensorflow_interpreter_wrapper.so
#18 0x0000ffffecbe7cf4 in pybind11::cpp_function::initialize<pybind11_init__pywrap_tensorflow_interpreter_wrapper(pybind11::module_&)::{lambda(tflite::interpreter_wrapper::InterpreterWrapper&, unsigned long)#24}, pybind11::obje--Type <RET> for more, q to quit, c to continue without paging--
ct, tflite::interpreter_wrapper::InterpreterWrapper&, unsigned long, pybind11::name, pybind11::is_method, pybind11::sibling, char [60]>(pybind11_init__pywrap_tensorflow_interpreter_wrapper(pybind11::module_&)::{lambda(tflite::interpreter_wrapper::InterpreterWrapper&, unsigned long)#24}&&, pybind11::object (*)(tflite::interpreter_wrapper::InterpreterWrapper&, unsigned long), pybind11::name const&, pybind11::is_method const&, pybind11::sibling const&, char const (&) [60])::{lambda(pybind11::detail::function_call&)#3}::_FUN(pybind11::detail::function_call&) () from /usr/lib/python3.10/site-packages/tflite_runtime/_pywrap_tensorflow_interpreter_wrapper.so
#19 0x0000ffffecbf5c50 in pybind11::cpp_function::dispatcher(_object*, _object*, _object*) () from /usr/lib/python3.10/site-packages/tflite_runtime/_pywrap_tensorflow_interpreter_wrapper.so
#20 0x0000fffff7d144fc in ?? () from /usr/lib/libpython3.10.so.1.0
#21 0x0000fffff7cd50cc in _PyObject_MakeTpCall () from /usr/lib/libpython3.10.so.1.0
#22 0x0000fffff7cd7e94 in ?? () from /usr/lib/libpython3.10.so.1.0
#23 0x0000fffff7c8a9d0 in _PyEval_EvalFrameDefault () from /usr/lib/libpython3.10.so.1.0
#24 0x0000fffff7d9eae8 in ?? () from /usr/lib/libpython3.10.so.1.0
#25 0x0000fffff7cd52e4 in _PyObject_FastCallDictTstate () from /usr/lib/libpython3.10.so.1.0
#26 0x0000fffff7cd5504 in _PyObject_Call_Prepend () from /usr/lib/libpython3.10.so.1.0
#27 0x0000fffff7d36430 in ?? () from /usr/lib/libpython3.10.so.1.0
#28 0x0000fffff7d2f7d0 in ?? () from /usr/lib/libpython3.10.so.1.0
#29 0x0000fffff7cd50cc in _PyObject_MakeTpCall () from /usr/lib/libpython3.10.so.1.0
#30 0x0000fffff7c8a6f8 in _PyEval_EvalFrameDefault () from /usr/lib/libpython3.10.so.1.0
#31 0x0000fffff7d9eae8 in ?? () from /usr/lib/libpython3.10.so.1.0
#32 0x0000fffff7c84d80 in _PyEval_EvalFrameDefault () from /usr/lib/libpython3.10.so.1.0
#33 0x0000fffff7d9eae8 in ?? () from /usr/lib/libpython3.10.so.1.0
#34 0x0000fffff7d9eca4 in PyEval_EvalCode () from /usr/lib/libpython3.10.so.1.0
#35 0x0000fffff7ddb810 in ?? () from /usr/lib/libpython3.10.so.1.0
#36 0x0000fffff7ddba38 in ?? () from /usr/lib/libpython3.10.so.1.0
#37 0x0000fffff7ddbb58 in ?? () from /usr/lib/libpython3.10.so.1.0
#38 0x0000fffff7ddd848 in _PyRun_SimpleFileObject () from /usr/lib/libpython3.10.so.1.0
#39 0x0000fffff7dddd1c in _PyRun_AnyFileObject () from /usr/lib/libpython3.10.so.1.0
#40 0x0000fffff7df9850 in Py_RunMain () from /usr/lib/libpython3.10.so.1.0
#41 0x0000fffff7df9f9c in Py_BytesMain () from /usr/lib/libpython3.10.so.1.0
#42 0x0000fffff7a9b230 in ?? () from /lib/libc.so.6
#43 0x0000fffff7a9b30c in __libc_start_main () from /lib/libc.so.6
#44 0x0000000000400870 in _start ()
(gdb)

Any ideas what is going on here?

Best Regards

0 Reese Grimsley 1 month ago in reply to Stefan Werner

TI__Genius 10636 points

Hi Stefan,

Thanks for the all the information here -- much appreciated and very helpful. I see the issue.

Stefan Werner said:
The EDGEAI_SDK_VERSION is set to 09_00_00. Since I've tried to update the target device to 10_00_07_00, I guess this is wrong, no

Unfortunately yes, this is probably an incompatible combination. Please see the version_compatibility doc. We started this form of backwards compatibility at 10.0 SDK and maintained compatibility (with the steps you found) for the 9.2 SDK. This does not apply for 9.0 SDK

So this is version compatibility issue. in doing this, you are applying 10.0.0.7 firmware that is compatible with 9.2 SDK in an actual 9.0 SDK installation.

Are you able to move SDK's to either 9.2 or 10.0? Worth noting that a 10.1 SDK will release within the next couple weeks. Otherwise, you would need to stick with edgeai-tidl-tools from 09_00_XX_YY

BR,
Reese

0 Dominic Mai 1 month ago in reply to Reese Grimsley

Prodigy 30 points

Hey Reese,

I'm a colleague of Stefan, we work on the same devboard (so all the information Stefan has given also holds for this post).

Reese Grimsley said:
Otherwise, you would need to stick with edgeai-tidl-tools from 09_00_XX_YY

I tried your suggestion to change the TIDL tools version in our devcontainer to 09_00_00_06

tidl-model-compilation/edgeai-tidl-tools$ git st
HEAD detached at 09_00_00_06

I think we also have some issues with updating the SDK on our devboard, Stefan tried to update to 10_00_07_00, but I think this was not successful (see above post for what Stefan tried.)

However, when I try the example compilation, the python script hangs (see error mdg after ctrl-c at the end)

root@c9f23fa83205:/opt/edgeai-tidl-tools/examples/osrt_python/tfl# python tflrt_delegate.py -c
Running 4 Models - ['cl-tfl-mobilenet_v1_1.0_224', 'ss-tfl-deeplabv3_mnv2_ade20k_float', 'od-tfl-ssd_mobilenet_v2_300_float', 'od-tfl-ssdlite_mobiledet_dsp_320x320_coco']


Running_Model :  cl-tfl-mobilenet_v1_1.0_224

Running_Model :  ss-tfl-deeplabv3_mnv2_ade20k_float

Running_Model :
Running_Model :  od-tfl-ssdlite_mobiledet_dsp_320x320_coco
 od-tfl-ssd_mobilenet_v2_300_float
Number of OD backbone nodes = 89
Size of odBackboneNodeIds = 89
TIDL Meta PipeLine (Proto) File  : ../../../models/public/ssdlite_mobiledet_dsp_320x320_coco_20200519.prototxt
Number of OD backbone nodes = 112
Size of odBackboneNodeIds = 112

 Preliminary number of subgraphs:1 , 81 nodes delegated out of 81 nodes


 Preliminary number of subgraphs:1 , 34 nodes delegated out of 34 nodes


 Preliminary number of subgraphs:1 , 129 nodes delegated out of 129 nodes

TF Meta PipeLine (Proto) File  : ../../../models/public/ssdlite_mobiledet_dsp_320x320_coco_20200519.prototxt
num_classes : 91
y_scale : 10.000000
x_scale : 10.000000
w_scale : 5.000000
h_scale : 5.000000
num_keypoints : 5.000000
score_threshold : 0.600000
iou_threshold : 0.450000
max_detections_per_class : 200
max_total_detections : 100
      scales, height_stride, width_stride, height_offset, width_offset
   0.2000000,   -1.0000000,   -1.0000000,   -1.0000000,   -1.0000000
   0.3500000,   -1.0000000,   -1.0000000,   -1.0000000,   -1.0000000
   0.5000000,   -1.0000000,   -1.0000000,   -1.0000000,   -1.0000000
   0.6500000,   -1.0000000,   -1.0000000,   -1.0000000,   -1.0000000
   0.8000000,   -1.0000000,   -1.0000000,   -1.0000000,   -1.0000000
   0.9500000,   -1.0000000,   -1.0000000,   -1.0000000,   -1.0000000
aspect_ratios
   1.0000000
   2.0000000
   0.5000000
   3.0000000
   0.3333000

 Preliminary number of subgraphs:1 , 107 nodes delegated out of 107 nodes

Warning : Requested Output Data Convert Layer is not Added to the network, It is currently not Optimal
Warning : Requested Output Data Convert Layer is not Added to the network, It is currently not Optimal

 ************** Frame index 1 : Running float import *************

 ************** Frame index 1 : Running float import *************
****************************************************
**                ALL MODEL CHECK PASSED          **
****************************************************

INFORMATION: [TIDL_ResizeLayer] ResizeBilinear_TIDL_0 Any resize ratio which is power of 2 and greater than 4 will be placed by combination of 4x4 resize layer and 2x2 resize layer. For example a 8x8 resize will be replaced by 4x4 resize followed by 2x2 resize.
INFORMATION: [TIDL_ResizeLayer] ResizeBilinear_TIDL_1 Any resize ratio which is power of 2 and greater than 4 will be placed by combination of 4x4 resize layer and 2x2 resize layer. For example a 8x8 resize will be replaced by 4x4 resize followed by 2x2 resize.
INFORMATION: [TIDL_ResizeLayer] ResizeBilinear Any resize ratio which is power of 2 and greater than 4 will be placed by combination of 4x4 resize layer and 2x2 resize layer. For example a 8x8 resize will be replaced by 4x4 resize followed by 2x2 resize.
INFORMATION: [TIDL_ResizeLayer] decoder/ResizeBilinear Any resize ratio which is power of 2 and greater than 4 will be placed by combination of 4x4 resize layer and 2x2 resize layer. For example a 8x8 resize will be replaced by 4x4 resize followed by 2x2 resize.
INFORMATION: [TIDL_ResizeLayer] ResizeBilinear_1 Any resize ratio which is power of 2 and greater than 4 will be placed by combination of 4x4 resize layer and 2x2 resize layer. For example a 8x8 resize will be replaced by 4x4 resize followed by 2x2 resize.
****************************************************
**          5 WARNINGS          0 ERRORS          **
****************************************************
The soft limit is 2048
The soft limit is 2048
The hard limit is 2048
The hard limit is 2048
MEM: Init ... !!!
MEM: Init ... !!!
MEM: Init ... Done !!!
MEM: Init ... Done !!!
Warning : Requested Output Data Convert Layer is not Added to the network, It is currently not Optimal
Warning : Requested Output Data Convert Layer is not Added to the network, It is currently not Optimal
Warning : Requested Output Data Convert Layer is not Added to the network, It is currently not Optimal
Warning : Requested Output Data Convert Layer is not Added to the network, It is currently not Optimal
 0.0s:  VX_ZONE_INIT:Enabled
 0.3s:  VX_ZONE_ERROR:Enabled
 0.4s:  VX_ZONE_WARNING:Enabled
 0.0s:  VX_ZONE_INIT:Enabled
 0.10s:  VX_ZONE_ERROR:Enabled
 0.11s:  VX_ZONE_WARNING:Enabled
 0.2903s:  VX_ZONE_INIT:[tivxInit:185] Initialization Done !!!
 0.3187s:  VX_ZONE_INIT:[tivxInit:185] Initialization Done !!!

 ************** Frame index 1 : Running float import *************
Warning : Requested Output Data Convert Layer is not Added to the network, It is currently not Optimal
Warning : Requested Output Data Convert Layer is not Added to the network, It is currently not Optimal
Warning : Requested Output Data Convert Layer is not Added to the network, It is currently not Optimal
Warning : Requested Output Data Convert Layer is not Added to the network, It is currently not Optimal

 ************** Frame index 1 : Running float import *************
****************************************************
**                ALL MODEL CHECK PASSED          **
****************************************************

The soft limit is 2048
The hard limit is 2048
MEM: Init ... !!!
MEM: Init ... Done !!!
 0.0s:  VX_ZONE_INIT:Enabled
 0.9s:  VX_ZONE_ERROR:Enabled
 0.10s:  VX_ZONE_WARNING:Enabled
 0.1993s:  VX_ZONE_INIT:[tivxInit:185] Initialization Done !!!
****************************************************
**                ALL MODEL CHECK PASSED          **
****************************************************

The soft limit is 2048
The hard limit is 2048
MEM: Init ... !!!
MEM: Init ... Done !!!
 0.0s:  VX_ZONE_INIT:Enabled
 0.12s:  VX_ZONE_ERROR:Enabled
 0.21s:  VX_ZONE_WARNING:Enabled
 0.2402s:  VX_ZONE_INIT:[tivxInit:185] Initialization Done !!!
 
 
 after CTRL-c:
 ^CTraceback (most recent call last):
  File "/opt/edgeai-tidl-tools/examples/osrt_python/tfl/tflrt_delegate.py", line 274, in <module>
    nthreads = join_one(nthreads)
  File "/opt/edgeai-tidl-tools/examples/osrt_python/tfl/tflrt_delegate.py", line 256, in join_one
    sem.acquire()
KeyboardInterrupt

This is what is being created:

root@c9f23fa83205:/opt/edgeai-tidl-tools/models/public# l
total 117M
8.7M -rw-r--r-- 1 root root 8.7M Dec 20 10:24 deeplabv3_mnv2_ade20k_float.tflite
 17M -rw-r--r-- 1 root root  17M Dec 20 10:24 mobilenet_v1_1.0_224.tflite
 28M -rw-r--r-- 1 root root  28M Dec 20 10:24 ssdlite_mobiledet_dsp_320x320_coco_20200519.tflite
4.0K -rw-r--r-- 1 root root 2.9K Dec 20 10:24 ssdlite_mobiledet_dsp_320x320_coco_20200519.prototxt
 65M -rw-r--r-- 1 root root  65M Dec 20 10:24 ssd_mobilenet_v2_300_float.tflite
 
 (3.10.16) root@c9f23fa83205:/opt/edgeai-tidl-tools/model-artifacts/cl-tfl-mobilenet_v1_1.0_224/tempDir# l
total 20M
 12K -rw-r--r-- 1 root root 8.8K Dec 20 11:43 86_tidl_net.bin_netLog.txt
 19M -rw-r--r-- 1 root root  19M Dec 20 11:43 86_tidl_net.bin
 40K -rw-r--r-- 1 root root  37K Dec 20 11:43 86_tidl_io_1.bin
4.0K -rw-r--r-- 1 root root 1.8K Dec 20 11:43 86_tidl_net.bin.layer_info.txt
236K -rw-r--r-- 1 root root 236K Dec 20 11:43 86_tidl_net.bin.svg

It runs locally, but no the devboard it tells me that "allowedNode.txt" is missing.

Any ideas what went wrong here?

(Note: not urgent, I'll be returning from christmas holidays mid january)

0 Reese Grimsley 1 month ago in reply to Dominic Mai

TI__Genius 10636 points

Hi Dominic,

Okay, so compiling against 9.0 SDK tools now, got it. This is correct if it is not feasible to upgrade the SDK otherwise.

Dominic Mai said:
However, when I try the example compilation, the python script hangs (see error mdg after ctrl-c at the end)

I see that you are running the default example here for the models. This will create fork multiple processes and may hang while it's waiting on one to return. Perhaps one of those failed. It is difficult to tell from the logs.

I noted on my side that the od-tfl-ssd_mobilenet_v2_300_float hit a segfault during compilation on my side.
- if one model doesn't complete and hangs, then the whole process will hang as the main thread waits for threads to complete.

Are you interested in a specific model or just trying to test the tools?

You can run a single model by adding '-m MODEL_CONFIG_NAME' to the command line args, where MODEL_CONFIG_NAME is a key from the examples/osrt_python/model_configs.py. One of these is "cl-tfl-mobilenet_v1_1.0_224".

The files in your tempDir look correct, but those are intermediate files (and some debugging info). The directory up from that has the important files for artifacts. There should be 2 binaries, a model file, and a few supporting files like this allowedNode.txt

I'd recommend increasing the debug_level parameter to 1. You can change this globally from the common_utils.py file or by adding 'debug_level': 1 to an additional "optional_options" dictionary within a model_configs.py dict entry. Most likely one model is failing and causing the whole script to hang.

BR,
Reese

0 Dominic Mai 16 days ago in reply to Reese Grimsley

Prodigy 30 points

Hi Reese,

I'm back from my holidays and tried out your suggestions - unfortunately none of them seem to have worked.

Reese Grimsley said:
Are you interested in a specific model or just trying to test the tools?

Currently, I'm only interested in completing the compilation / deployment workflow for an arbitrary model. Next would be to deploy a custom model and do evaluations (accuracy, inference time) with it.

Reese Grimsley said:
You can run a single model by adding '-m MODEL_CONFIG_NAME' to the command line args, where MODEL_CONFIG_NAME is a key from the examples/osrt_python/model_configs.py. One of these is "cl-tfl-mobilenet_v1_1.0_224".

I think the `-m` option is not available (yet?) in the script `root@cd64c3ba8cc1:/opt/edgeai-tidl-tools/examples/osrt_python/tfl# python tflrt_delegate.py -c` I'm calling. But I just manually edited the `models` list in line 240.

I also set `ncpus = 1` (line 41), otherwise I still get the thread related error (os.cpu_count() == 24 on my system):

^CTraceback (most recent call last):
  File "/opt/edgeai-tidl-tools/examples/osrt_python/tfl/tflrt_delegate.py", line 275, in <module>
    nthreads = join_one(nthreads)
  File "/opt/edgeai-tidl-tools/examples/osrt_python/tfl/tflrt_delegate.py", line 257, in join_one
    sem.acquire()
KeyboardInterrupt

I tried all the following models with the same result

'cl-tfl-mobilenet_v1_1.0_224'
'ss-tfl-deeplabv3_mnv2_ade20k_float'
'od-tfl-ssd_mobilenet_v2_300_float'
'od-tfl-ssdlite_mobiledet_dsp_320x320_coco'

Exemplary output with debug level 1 and ncpus = 1

(3.10.16) root@cd64c3ba8cc1:/opt/edgeai-tidl-tools/examples/osrt_python/tfl# python tflrt_delegate.py -c
Running 1 Models - ['od-tfl-ssdlite_mobiledet_dsp_320x320_coco']


Running_Model :  od-tfl-ssdlite_mobiledet_dsp_320x320_coco
tidl_tools_path                                 = /opt/edgeai-tidl-tools/tidl_tools 
artifacts_folder                                = ../../../model-artifacts//od-tfl-ssdlite_mobiledet_dsp_320x320_coco/ 
tidl_tensor_bits                                = 8 
debug_level                                     = 1 
num_tidl_subgraphs                              = 16 
tidl_denylist                                   = 
tidl_denylist_layer_name                        = 
tidl_denylist_layer_type                         = 
tidl_allowlist_layer_name                        = 
model_type                                      =  
tidl_calibration_accuracy_level                 = 7 
tidl_calibration_options:num_frames_calibration = 2 
tidl_calibration_options:bias_calibration_iterations = 5 
mixed_precision_factor = -1.000000 
model_group_id = 0 
power_of_2_quantization                         = 2 
enable_high_resolution_optimization             = 0 
pre_batchnorm_fold                              = 1 
add_data_convert_ops                          = 3 
output_feature_16bit_names_list                 =  
m_params_16bit_names_list                       =  
reserved_compile_constraints_flag               = 1601 
ti_internal_reserved_1                          = 
TIDL Meta PipeLine (Proto) File  : ../../../models/public/ssdlite_mobiledet_dsp_320x320_coco_20200519.prototxt  
Number of OD backbone nodes = 112 
Size of odBackboneNodeIds = 112 
Supported TIDL layer type --- 54 Tflite layer type --- 53 layer output name--- FeatureExtractor/MobileDetDSP/Conv/Relu6;FeatureExtractor/MobileDetDSP/Conv/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBNNoExpansion/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBNNoExpansion/SeparableConv2d/depthwise;FeatureExtractor/MobileDetDSP/TuckerConv/Conv_2/Conv2D;FeatureExtractor/MobileDetDSP/Conv/Conv2D/Mul/Bias/InCast 
Supported TIDL layer type --- 8 Tflite layer type --- 0 layer output name--- FeatureExtractor/MobileDetDSP/Conv/Relu6;FeatureExtractor/MobileDetDSP/Conv/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBNNoExpansion/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBNNoExpansion/SeparableConv2d/depthwise;FeatureExtractor/MobileDetDSP/TuckerConv/Conv_2/Conv2D;FeatureExtractor/MobileDetDSP/Conv/Conv2D/Mul/Bias 
Supported TIDL layer type --- 8 Tflite layer type --- 18 layer output name--- FeatureExtractor/MobileDetDSP/Conv/Relu6;FeatureExtractor/MobileDetDSP/Conv/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBNNoExpansion/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBNNoExpansion/SeparableConv2d/depthwise;FeatureExtractor/MobileDetDSP/TuckerConv/Conv_2/Conv2D;FeatureExtractor/MobileDetDSP/Conv/Conv2D/Mul 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/Conv/Relu6;FeatureExtractor/MobileDetDSP/Conv/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBNNoExpansion/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBNNoExpansion/SeparableConv2d/depthwise;FeatureExtractor/MobileDetDSP/TuckerConv/Conv_2/Conv2D;FeatureExtractor/MobileDetDSP/Conv/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- FeatureExtractor/MobileDetDSP/IBNNoExpansion/SeparableConv2d/Relu6;FeatureExtractor/MobileDetDSP/IBNNoExpansion/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBNNoExpansion/SeparableConv2d/depthwise;FeatureExtractor/MobileDetDSP/TuckerConv/Conv_2/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/IBNNoExpansion/Conv/BatchNorm/FusedBatchNormV3;BoxPredictor_5/BoxEncodingPredictor/Conv2D;FeatureExtractor/MobileDetDSP/IBNNoExpansion/Conv/Conv2D1 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/FusedConv/Conv/Relu6;FeatureExtractor/MobileDetDSP/FusedConv/Conv/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/FusedConv/Conv/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/FusedConv/Conv_1/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/TuckerConv/Conv_2/Conv2D;FeatureExtractor/MobileDetDSP/FusedConv/Conv_1/Conv2D1 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/FusedConv_1/Conv/Relu6;FeatureExtractor/MobileDetDSP/FusedConv_1/Conv/BatchNorm/FusedBatchNormV3;BoxPredictor_5/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_5/ClassPredictor_depthwise/depthwise;FeatureExtractor/MobileDetDSP/FusedConv_1/Conv/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/FusedConv_1/Conv_1/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/TuckerConv/Conv_2/Conv2D;FeatureExtractor/MobileDetDSP/FusedConv_1/Conv_1/Conv2D1 
Supported TIDL layer type --- 5 Tflite layer type --- 0 layer output name--- FeatureExtractor/MobileDetDSP/FusedConv_1/add 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/IBN/Conv/Relu6;FeatureExtractor/MobileDetDSP/IBN/Conv/BatchNorm/FusedBatchNormV3;BoxPredictor_5/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_5/ClassPredictor_depthwise/depthwise;FeatureExtractor/MobileDetDSP/IBN/Conv/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- FeatureExtractor/MobileDetDSP/IBN/SeparableConv2d/Relu6;FeatureExtractor/MobileDetDSP/IBN/SeparableConv2d/BatchNorm/FusedBatchNormV3;BoxPredictor_5/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_5/ClassPredictor_depthwise/depthwise;FeatureExtractor/MobileDetDSP/IBN/SeparableConv2d/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/IBN/Conv_1/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/TuckerConv/Conv_2/Conv2D;FeatureExtractor/MobileDetDSP/IBN/Conv_1/Conv2D1 
Supported TIDL layer type --- 5 Tflite layer type --- 0 layer output name--- FeatureExtractor/MobileDetDSP/IBN/add 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/TuckerConv/Conv/Relu6;FeatureExtractor/MobileDetDSP/TuckerConv/Conv/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/TuckerConv/Conv/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/TuckerConv/Conv_1/Relu6;FeatureExtractor/MobileDetDSP/TuckerConv/Conv_1/BatchNorm/FusedBatchNormV3;BoxPredictor_5/BoxEncodingPredictor/Conv2D;FeatureExtractor/MobileDetDSP/TuckerConv/Conv_1/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/TuckerConv/Conv_2/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/TuckerConv/Conv_2/Conv2D1 
Supported TIDL layer type --- 5 Tflite layer type --- 0 layer output name--- FeatureExtractor/MobileDetDSP/TuckerConv/add 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/FusedConv_2/Conv/Relu6;FeatureExtractor/MobileDetDSP/FusedConv_2/Conv/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/depthwise;FeatureExtractor/MobileDetDSP/FusedConv_2/Conv/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/FusedConv_2/Conv_1/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/C5_2_Conv2d_5_3x3_s2_128_depthwise/depthwise;FeatureExtractor/MobileDetDSP/FusedConv_2/Conv_1/Conv2D1 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/IBN_1/Conv/Relu6;FeatureExtractor/MobileDetDSP/IBN_1/Conv/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/depthwise;FeatureExtractor/MobileDetDSP/IBN_1/Conv/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- FeatureExtractor/MobileDetDSP/IBN_1/SeparableConv2d/Relu6;FeatureExtractor/MobileDetDSP/IBN_1/SeparableConv2d/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/depthwise;FeatureExtractor/MobileDetDSP/IBN_1/SeparableConv2d/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/IBN_1/Conv_1/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/C5_2_Conv2d_5_3x3_s2_128_depthwise/depthwise;FeatureExtractor/MobileDetDSP/IBN_1/Conv_1/Conv2D1 
Supported TIDL layer type --- 5 Tflite layer type --- 0 layer output name--- FeatureExtractor/MobileDetDSP/IBN_1/add 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/FusedConv_3/Conv/Relu6;FeatureExtractor/MobileDetDSP/FusedConv_3/Conv/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/depthwise;FeatureExtractor/MobileDetDSP/FusedConv_3/Conv/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/FusedConv_3/Conv_1/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/C5_2_Conv2d_5_3x3_s2_128_depthwise/depthwise;FeatureExtractor/MobileDetDSP/FusedConv_3/Conv_1/Conv2D1 
Supported TIDL layer type --- 5 Tflite layer type --- 0 layer output name--- FeatureExtractor/MobileDetDSP/FusedConv_3/add 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/FusedConv_4/Conv/Relu6;FeatureExtractor/MobileDetDSP/FusedConv_4/Conv/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/depthwise;FeatureExtractor/MobileDetDSP/FusedConv_4/Conv/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/FusedConv_4/Conv_1/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/C5_2_Conv2d_5_3x3_s2_128_depthwise/depthwise;FeatureExtractor/MobileDetDSP/FusedConv_4/Conv_1/Conv2D1 
Supported TIDL layer type --- 5 Tflite layer type --- 0 layer output name--- FeatureExtractor/MobileDetDSP/FusedConv_4/add 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/FusedConv_5/Conv/Relu6;FeatureExtractor/MobileDetDSP/FusedConv_5/Conv/BatchNorm/FusedBatchNormV3;BoxPredictor_2/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_2/ClassPredictor_depthwise/depthwise;FeatureExtractor/MobileDetDSP/FusedConv_5/Conv/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/FusedConv_5/Conv_1/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/TuckerConv_1/Conv_1/Conv2D;FeatureExtractor/MobileDetDSP/FusedConv_5/Conv_1/Conv2D1 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/IBN_2/Conv/Relu6;FeatureExtractor/MobileDetDSP/IBN_2/Conv/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_2/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_2/SeparableConv2d/depthwise;FeatureExtractor/MobileDetDSP/IBN_2/Conv/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- FeatureExtractor/MobileDetDSP/IBN_2/SeparableConv2d/Relu6;FeatureExtractor/MobileDetDSP/IBN_2/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_2/SeparableConv2d/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/IBN_2/Conv_1/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/TuckerConv_1/Conv_1/Conv2D;FeatureExtractor/MobileDetDSP/IBN_2/Conv_1/Conv2D1 
Supported TIDL layer type --- 5 Tflite layer type --- 0 layer output name--- FeatureExtractor/MobileDetDSP/IBN_2/add 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/IBN_3/Conv/Relu6;FeatureExtractor/MobileDetDSP/IBN_3/Conv/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_4/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_4/SeparableConv2d/depthwise;FeatureExtractor/MobileDetDSP/FusedConv_6/Conv/Conv2D;FeatureExtractor/MobileDetDSP/IBN_3/Conv/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- FeatureExtractor/MobileDetDSP/IBN_3/SeparableConv2d/Relu6;FeatureExtractor/MobileDetDSP/IBN_3/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_4/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_4/SeparableConv2d/depthwise;FeatureExtractor/MobileDetDSP/FusedConv_6/Conv/Conv2D;FeatureExtractor/MobileDetDSP/IBN_3/SeparableConv2d/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/IBN_3/Conv_1/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/TuckerConv_1/Conv_1/Conv2D;FeatureExtractor/MobileDetDSP/IBN_3/Conv_1/Conv2D1 
Supported TIDL layer type --- 5 Tflite layer type --- 0 layer output name--- FeatureExtractor/MobileDetDSP/IBN_3/add 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/IBN_4/Conv/Relu6;FeatureExtractor/MobileDetDSP/IBN_4/Conv/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_4/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_4/SeparableConv2d/depthwise;FeatureExtractor/MobileDetDSP/FusedConv_6/Conv/Conv2D;FeatureExtractor/MobileDetDSP/IBN_4/Conv/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- FeatureExtractor/MobileDetDSP/IBN_4/SeparableConv2d/Relu6;FeatureExtractor/MobileDetDSP/IBN_4/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_4/SeparableConv2d/depthwise;FeatureExtractor/MobileDetDSP/FusedConv_6/Conv/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/IBN_4/Conv_1/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/TuckerConv_1/Conv_1/Conv2D;FeatureExtractor/MobileDetDSP/IBN_4/Conv_1/Conv2D1 
Supported TIDL layer type --- 5 Tflite layer type --- 0 layer output name--- FeatureExtractor/MobileDetDSP/IBN_4/add 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/FusedConv_6/Conv/Relu6;FeatureExtractor/MobileDetDSP/FusedConv_6/Conv/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_4/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_4/SeparableConv2d/depthwise;FeatureExtractor/MobileDetDSP/FusedConv_6/Conv/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/FusedConv_6/Conv_1/BatchNorm/FusedBatchNormV3;BoxPredictor_0/ClassPredictor_depthwise/depthwise;FeatureExtractor/MobileDetDSP/FusedConv_6/Conv_1/Conv2D1 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/IBN_5/Conv/Relu6;FeatureExtractor/MobileDetDSP/IBN_5/Conv/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_7/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_7/SeparableConv2d/depthwise;FeatureExtractor/MobileDetDSP/IBN_5/Conv/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- FeatureExtractor/MobileDetDSP/IBN_5/SeparableConv2d/Relu6;FeatureExtractor/MobileDetDSP/IBN_5/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_7/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_7/SeparableConv2d/depthwise;FeatureExtractor/MobileDetDSP/IBN_5/SeparableConv2d/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/IBN_5/Conv_1/BatchNorm/FusedBatchNormV3;BoxPredictor_0/ClassPredictor_depthwise/depthwise;FeatureExtractor/MobileDetDSP/IBN_5/Conv_1/Conv2D1 
Supported TIDL layer type --- 5 Tflite layer type --- 0 layer output name--- FeatureExtractor/MobileDetDSP/IBN_5/add 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/IBN_6/Conv/Relu6;FeatureExtractor/MobileDetDSP/IBN_6/Conv/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_7/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_7/SeparableConv2d/depthwise;FeatureExtractor/MobileDetDSP/IBN_6/Conv/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- FeatureExtractor/MobileDetDSP/IBN_6/SeparableConv2d/Relu6;FeatureExtractor/MobileDetDSP/IBN_6/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_7/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_7/SeparableConv2d/depthwise;FeatureExtractor/MobileDetDSP/IBN_6/SeparableConv2d/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/IBN_6/Conv_1/BatchNorm/FusedBatchNormV3;BoxPredictor_0/ClassPredictor_depthwise/depthwise;FeatureExtractor/MobileDetDSP/IBN_6/Conv_1/Conv2D1 
Supported TIDL layer type --- 5 Tflite layer type --- 0 layer output name--- FeatureExtractor/MobileDetDSP/IBN_6/add 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/IBN_7/Conv/Relu6;FeatureExtractor/MobileDetDSP/IBN_7/Conv/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_7/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_7/SeparableConv2d/depthwise;FeatureExtractor/MobileDetDSP/IBN_7/Conv/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- FeatureExtractor/MobileDetDSP/IBN_7/SeparableConv2d/Relu6;FeatureExtractor/MobileDetDSP/IBN_7/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_7/SeparableConv2d/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/IBN_7/Conv_1/BatchNorm/FusedBatchNormV3;BoxPredictor_0/ClassPredictor_depthwise/depthwise;FeatureExtractor/MobileDetDSP/IBN_7/Conv_1/Conv2D1 
Supported TIDL layer type --- 5 Tflite layer type --- 0 layer output name--- FeatureExtractor/MobileDetDSP/IBN_7/add 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- BoxPredictor_0/BoxEncodingPredictor_depthwise/Relu6;BoxPredictor_0/BoxEncodingPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_0/BoxEncodingPredictor_depthwise/depthwise;BoxPredictor_0/ClassPredictor_depthwise/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- BoxPredictor_0/BoxEncodingPredictor/BiasAdd;BoxPredictor_0/BoxEncodingPredictor/Conv2D;BoxPredictor_0/BoxEncodingPredictor/biases1 
Supported TIDL layer type --- 0 Tflite layer type --- 22 layer output name--- BoxPredictor_0/Reshape 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- BoxPredictor_0/ClassPredictor_depthwise/Relu6;BoxPredictor_0/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_0/ClassPredictor_depthwise/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- BoxPredictor_0/ClassPredictor/BiasAdd;BoxPredictor_0/ClassPredictor/Conv2D;BoxPredictor_0/ClassPredictor/biases1 
Supported TIDL layer type --- 0 Tflite layer type --- 22 layer output name--- BoxPredictor_0/Reshape_1 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/IBN_8/Conv/Relu6;FeatureExtractor/MobileDetDSP/IBN_8/Conv/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_8/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_8/SeparableConv2d/depthwise;FeatureExtractor/MobileDetDSP/IBN_8/Conv/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- FeatureExtractor/MobileDetDSP/IBN_8/SeparableConv2d/Relu6;FeatureExtractor/MobileDetDSP/IBN_8/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_8/SeparableConv2d/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/IBN_8/Conv_1/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/TuckerConv_1/Conv_2/Conv2D;FeatureExtractor/MobileDetDSP/IBN_8/Conv_1/Conv2D1 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/IBN_9/Conv/Relu6;FeatureExtractor/MobileDetDSP/IBN_9/Conv/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_9/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_9/SeparableConv2d/depthwise;FeatureExtractor/MobileDetDSP/FusedConv_7/Conv/Conv2D;FeatureExtractor/MobileDetDSP/IBN_9/Conv/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- FeatureExtractor/MobileDetDSP/IBN_9/SeparableConv2d/Relu6;FeatureExtractor/MobileDetDSP/IBN_9/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_9/SeparableConv2d/depthwise;FeatureExtractor/MobileDetDSP/FusedConv_7/Conv/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/IBN_9/Conv_1/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/TuckerConv_1/Conv_2/Conv2D;FeatureExtractor/MobileDetDSP/IBN_9/Conv_1/Conv2D1 
Supported TIDL layer type --- 5 Tflite layer type --- 0 layer output name--- FeatureExtractor/MobileDetDSP/IBN_9/add 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/FusedConv_7/Conv/Relu6;FeatureExtractor/MobileDetDSP/FusedConv_7/Conv/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_9/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_9/SeparableConv2d/depthwise;FeatureExtractor/MobileDetDSP/FusedConv_7/Conv/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/FusedConv_7/Conv_1/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/TuckerConv_1/Conv_2/Conv2D;FeatureExtractor/MobileDetDSP/FusedConv_7/Conv_1/Conv2D1 
Supported TIDL layer type --- 5 Tflite layer type --- 0 layer output name--- FeatureExtractor/MobileDetDSP/FusedConv_7/add 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/TuckerConv_1/Conv/Relu6;FeatureExtractor/MobileDetDSP/TuckerConv_1/Conv/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/TuckerConv_1/Conv_1/Conv2D;FeatureExtractor/MobileDetDSP/TuckerConv_1/Conv/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/TuckerConv_1/Conv_1/Relu6;FeatureExtractor/MobileDetDSP/TuckerConv_1/Conv_1/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/TuckerConv_1/Conv_1/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/TuckerConv_1/Conv_2/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/TuckerConv_1/Conv_2/Conv2D1 
Supported TIDL layer type --- 5 Tflite layer type --- 0 layer output name--- FeatureExtractor/MobileDetDSP/TuckerConv_1/add 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/IBN_10/Conv/Relu6;FeatureExtractor/MobileDetDSP/IBN_10/Conv/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_10/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_10/SeparableConv2d/depthwise;FeatureExtractor/MobileDetDSP/IBN_10/Conv/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- FeatureExtractor/MobileDetDSP/IBN_10/SeparableConv2d/Relu6;FeatureExtractor/MobileDetDSP/IBN_10/SeparableConv2d/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/IBN_10/SeparableConv2d/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/IBN_10/Conv_1/BatchNorm/FusedBatchNormV3;BoxPredictor_1/ClassPredictor_depthwise/depthwise;FeatureExtractor/MobileDetDSP/IBN_10/Conv_1/Conv2D1 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- BoxPredictor_1/BoxEncodingPredictor_depthwise/Relu6;BoxPredictor_1/BoxEncodingPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_1/BoxEncodingPredictor_depthwise/depthwise;BoxPredictor_1/ClassPredictor_depthwise/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- BoxPredictor_1/BoxEncodingPredictor/BiasAdd;BoxPredictor_5/BoxEncodingPredictor/Conv2D;BoxPredictor_1/BoxEncodingPredictor/Conv2D;BoxPredictor_1/BoxEncodingPredictor/biases1 
Supported TIDL layer type --- 0 Tflite layer type --- 22 layer output name--- BoxPredictor_1/Reshape 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- BoxPredictor_1/ClassPredictor_depthwise/Relu6;BoxPredictor_1/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_1/ClassPredictor_depthwise/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- BoxPredictor_1/ClassPredictor/BiasAdd;BoxPredictor_5/ClassPredictor/Conv2D;BoxPredictor_1/ClassPredictor/Conv2D;BoxPredictor_1/ClassPredictor/biases1 
Supported TIDL layer type --- 0 Tflite layer type --- 22 layer output name--- BoxPredictor_1/Reshape_1 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/C5_1_Conv2d_2_1x1_256/Relu6;FeatureExtractor/MobileDetDSP/C5_1_Conv2d_2_1x1_256/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/depthwise;FeatureExtractor/MobileDetDSP/C5_1_Conv2d_2_1x1_256/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- FeatureExtractor/MobileDetDSP/C5_2_Conv2d_2_3x3_s2_512_depthwise/Relu6;FeatureExtractor/MobileDetDSP/C5_2_Conv2d_2_3x3_s2_512_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/depthwise;FeatureExtractor/MobileDetDSP/C5_2_Conv2d_2_3x3_s2_512_depthwise/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/C5_2_Conv2d_2_3x3_s2_512/Relu6;FeatureExtractor/MobileDetDSP/C5_2_Conv2d_2_3x3_s2_512/BatchNorm/FusedBatchNormV3;BoxPredictor_2/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_2/ClassPredictor_depthwise/depthwise;FeatureExtractor/MobileDetDSP/C5_2_Conv2d_2_3x3_s2_512/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- BoxPredictor_2/BoxEncodingPredictor_depthwise/Relu6;BoxPredictor_2/BoxEncodingPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_2/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_2/ClassPredictor_depthwise/depthwise;BoxPredictor_2/BoxEncodingPredictor_depthwise/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- BoxPredictor_2/BoxEncodingPredictor/BiasAdd;BoxPredictor_5/BoxEncodingPredictor/Conv2D;BoxPredictor_2/BoxEncodingPredictor/Conv2D;BoxPredictor_2/BoxEncodingPredictor/biases1 
Supported TIDL layer type --- 0 Tflite layer type --- 22 layer output name--- BoxPredictor_2/Reshape 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- BoxPredictor_2/ClassPredictor_depthwise/Relu6;BoxPredictor_2/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_2/ClassPredictor_depthwise/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- BoxPredictor_2/ClassPredictor/BiasAdd;BoxPredictor_5/ClassPredictor/Conv2D;BoxPredictor_2/ClassPredictor/Conv2D;BoxPredictor_2/ClassPredictor/biases1 
Supported TIDL layer type --- 0 Tflite layer type --- 22 layer output name--- BoxPredictor_2/Reshape_1 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/C5_1_Conv2d_3_1x1_128/Relu6;FeatureExtractor/MobileDetDSP/C5_1_Conv2d_3_1x1_128/BatchNorm/FusedBatchNormV3;BoxPredictor_5/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_5/ClassPredictor_depthwise/depthwise;FeatureExtractor/MobileDetDSP/C5_1_Conv2d_3_1x1_128/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- FeatureExtractor/MobileDetDSP/C5_2_Conv2d_3_3x3_s2_256_depthwise/Relu6;FeatureExtractor/MobileDetDSP/C5_2_Conv2d_3_3x3_s2_256_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_5/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_5/ClassPredictor_depthwise/depthwise;FeatureExtractor/MobileDetDSP/C5_2_Conv2d_3_3x3_s2_256_depthwise/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/C5_2_Conv2d_3_3x3_s2_256/Relu6;FeatureExtractor/MobileDetDSP/C5_2_Conv2d_3_3x3_s2_256/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/depthwise;FeatureExtractor/MobileDetDSP/C5_2_Conv2d_3_3x3_s2_256/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- BoxPredictor_3/BoxEncodingPredictor_depthwise/Relu6;BoxPredictor_3/BoxEncodingPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/depthwise;BoxPredictor_3/BoxEncodingPredictor_depthwise/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- BoxPredictor_3/BoxEncodingPredictor/BiasAdd;BoxPredictor_5/BoxEncodingPredictor/Conv2D;BoxPredictor_3/BoxEncodingPredictor/Conv2D;BoxPredictor_3/BoxEncodingPredictor/biases1 
Supported TIDL layer type --- 0 Tflite layer type --- 22 layer output name--- BoxPredictor_3/Reshape 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- BoxPredictor_3/ClassPredictor_depthwise/Relu6;BoxPredictor_3/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/depthwise;BoxPredictor_3/ClassPredictor_depthwise/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- BoxPredictor_3/ClassPredictor/BiasAdd;BoxPredictor_5/ClassPredictor/Conv2D;BoxPredictor_3/ClassPredictor/Conv2D;BoxPredictor_3/ClassPredictor/biases1 
Supported TIDL layer type --- 0 Tflite layer type --- 22 layer output name--- BoxPredictor_3/Reshape_1 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/C5_1_Conv2d_4_1x1_128/Relu6;FeatureExtractor/MobileDetDSP/C5_1_Conv2d_4_1x1_128/BatchNorm/FusedBatchNormV3;BoxPredictor_5/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_5/ClassPredictor_depthwise/depthwise;FeatureExtractor/MobileDetDSP/C5_1_Conv2d_4_1x1_128/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- FeatureExtractor/MobileDetDSP/C5_2_Conv2d_4_3x3_s2_256_depthwise/Relu6;FeatureExtractor/MobileDetDSP/C5_2_Conv2d_4_3x3_s2_256_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_5/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_5/ClassPredictor_depthwise/depthwise;FeatureExtractor/MobileDetDSP/C5_2_Conv2d_4_3x3_s2_256_depthwise/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/C5_2_Conv2d_4_3x3_s2_256/Relu6;FeatureExtractor/MobileDetDSP/C5_2_Conv2d_4_3x3_s2_256/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/depthwise;FeatureExtractor/MobileDetDSP/C5_2_Conv2d_4_3x3_s2_256/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- BoxPredictor_4/BoxEncodingPredictor_depthwise/Relu6;BoxPredictor_4/BoxEncodingPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/depthwise;BoxPredictor_4/BoxEncodingPredictor_depthwise/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- BoxPredictor_4/BoxEncodingPredictor/BiasAdd;BoxPredictor_5/BoxEncodingPredictor/Conv2D;BoxPredictor_4/BoxEncodingPredictor/Conv2D;BoxPredictor_4/BoxEncodingPredictor/biases1 
Supported TIDL layer type --- 0 Tflite layer type --- 22 layer output name--- BoxPredictor_4/Reshape 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- BoxPredictor_4/ClassPredictor_depthwise/Relu6;BoxPredictor_4/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_4/ClassPredictor_depthwise/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- BoxPredictor_4/ClassPredictor/BiasAdd;BoxPredictor_5/ClassPredictor/Conv2D;BoxPredictor_4/ClassPredictor/Conv2D;BoxPredictor_4/ClassPredictor/biases1 
Supported TIDL layer type --- 0 Tflite layer type --- 22 layer output name--- BoxPredictor_4/Reshape_1 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/C5_1_Conv2d_5_1x1_64/Relu6;FeatureExtractor/MobileDetDSP/C5_1_Conv2d_5_1x1_64/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/C5_2_Conv2d_5_3x3_s2_128_depthwise/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/C5_2_Conv2d_5_3x3_s2_128_depthwise/depthwise;FeatureExtractor/MobileDetDSP/C5_1_Conv2d_5_1x1_64/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- FeatureExtractor/MobileDetDSP/C5_2_Conv2d_5_3x3_s2_128_depthwise/Relu6;FeatureExtractor/MobileDetDSP/C5_2_Conv2d_5_3x3_s2_128_depthwise/BatchNorm/FusedBatchNormV3;FeatureExtractor/MobileDetDSP/C5_2_Conv2d_5_3x3_s2_128_depthwise/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- FeatureExtractor/MobileDetDSP/C5_2_Conv2d_5_3x3_s2_128/Relu6;FeatureExtractor/MobileDetDSP/C5_2_Conv2d_5_3x3_s2_128/BatchNorm/FusedBatchNormV3;BoxPredictor_5/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_5/ClassPredictor_depthwise/depthwise;FeatureExtractor/MobileDetDSP/C5_2_Conv2d_5_3x3_s2_128/Conv2D 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- BoxPredictor_5/BoxEncodingPredictor_depthwise/Relu6;BoxPredictor_5/BoxEncodingPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_5/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_5/ClassPredictor_depthwise/depthwise;BoxPredictor_5/BoxEncodingPredictor_depthwise/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- BoxPredictor_5/BoxEncodingPredictor/BiasAdd;BoxPredictor_5/BoxEncodingPredictor/Conv2D;BoxPredictor_5/BoxEncodingPredictor/biases1 
Supported TIDL layer type --- 0 Tflite layer type --- 22 layer output name--- BoxPredictor_5/Reshape 
Supported TIDL layer type --- 0 Tflite layer type --- 2 layer output name---        concat_1 
Supported TIDL layer type --- 0 Tflite layer type --- 22 layer output name---        Squeeze1 
Supported TIDL layer type --- 1 Tflite layer type --- 4 layer output name--- BoxPredictor_5/ClassPredictor_depthwise/Relu6;BoxPredictor_5/ClassPredictor_depthwise/BatchNorm/FusedBatchNormV3;BoxPredictor_5/ClassPredictor_depthwise/depthwise 
Supported TIDL layer type --- 1 Tflite layer type --- 3 layer output name--- BoxPredictor_5/ClassPredictor/BiasAdd;BoxPredictor_5/ClassPredictor/Conv2D;BoxPredictor_5/ClassPredictor/biases1 
Supported TIDL layer type --- 0 Tflite layer type --- 22 layer output name--- BoxPredictor_5/Reshape_1 
Supported TIDL layer type --- 0 Tflite layer type --- 2 layer output name---          concat 
Supported TIDL layer type --- 0 Tflite layer type --- 14 layer output name---  convert_scores 
Supported TIDL layer type --- 0 Tflite layer type --- 32 layer output name--- TFLite_Detection_PostProcess 

 Preliminary number of subgraphs:1 , 129 nodes delegated out of 129 nodes 
 
In TIDL_tfliteRtImportInit subgraph_id=321
Layer 0, subgraph id 321, name=BoxPredictor_0/BoxEncodingPredictor/BiasAdd;BoxPredictor_0/BoxEncodingPredictor/Conv2D;BoxPredictor_0/BoxEncodingPredictor/biases1
Layer 1, subgraph id 321, name=BoxPredictor_0/ClassPredictor/BiasAdd;BoxPredictor_0/ClassPredictor/Conv2D;BoxPredictor_0/ClassPredictor/biases1
Layer 2, subgraph id 321, name=BoxPredictor_1/BoxEncodingPredictor/BiasAdd;BoxPredictor_5/BoxEncodingPredictor/Conv2D;BoxPredictor_1/BoxEncodingPredictor/Conv2D;BoxPredictor_1/BoxEncodingPredictor/biases1
Layer 3, subgraph id 321, name=BoxPredictor_1/ClassPredictor/BiasAdd;BoxPredictor_5/ClassPredictor/Conv2D;BoxPredictor_1/ClassPredictor/Conv2D;BoxPredictor_1/ClassPredictor/biases1
Layer 4, subgraph id 321, name=BoxPredictor_2/BoxEncodingPredictor/BiasAdd;BoxPredictor_5/BoxEncodingPredictor/Conv2D;BoxPredictor_2/BoxEncodingPredictor/Conv2D;BoxPredictor_2/BoxEncodingPredictor/biases1
Layer 5, subgraph id 321, name=BoxPredictor_2/ClassPredictor/BiasAdd;BoxPredictor_5/ClassPredictor/Conv2D;BoxPredictor_2/ClassPredictor/Conv2D;BoxPredictor_2/ClassPredictor/biases1
Layer 6, subgraph id 321, name=BoxPredictor_3/BoxEncodingPredictor/BiasAdd;BoxPredictor_5/BoxEncodingPredictor/Conv2D;BoxPredictor_3/BoxEncodingPredictor/Conv2D;BoxPredictor_3/BoxEncodingPredictor/biases1
Layer 7, subgraph id 321, name=BoxPredictor_3/ClassPredictor/BiasAdd;BoxPredictor_5/ClassPredictor/Conv2D;BoxPredictor_3/ClassPredictor/Conv2D;BoxPredictor_3/ClassPredictor/biases1
Layer 8, subgraph id 321, name=BoxPredictor_4/BoxEncodingPredictor/BiasAdd;BoxPredictor_5/BoxEncodingPredictor/Conv2D;BoxPredictor_4/BoxEncodingPredictor/Conv2D;BoxPredictor_4/BoxEncodingPredictor/biases1
Layer 9, subgraph id 321, name=BoxPredictor_4/ClassPredictor/BiasAdd;BoxPredictor_5/ClassPredictor/Conv2D;BoxPredictor_4/ClassPredictor/Conv2D;BoxPredictor_4/ClassPredictor/biases1
Layer 10, subgraph id 321, name=BoxPredictor_5/BoxEncodingPredictor/BiasAdd;BoxPredictor_5/BoxEncodingPredictor/Conv2D;BoxPredictor_5/BoxEncodingPredictor/biases1
Layer 11, subgraph id 321, name=BoxPredictor_5/ClassPredictor/BiasAdd;BoxPredictor_5/ClassPredictor/Conv2D;BoxPredictor_5/ClassPredictor/biases1
Layer 12, subgraph id 321, name=normalized_input_image_tensor
TF Meta PipeLine (Proto) File  : ../../../models/public/ssdlite_mobiledet_dsp_320x320_coco_20200519.prototxt  
num_classes : 91
y_scale : 10.000000
x_scale : 10.000000
w_scale : 5.000000
h_scale : 5.000000
num_keypoints : 5.000000
score_threshold : 0.600000
iou_threshold : 0.450000
max_detections_per_class : 200
max_total_detections : 100
      scales, height_stride, width_stride, height_offset, width_offset
   0.2000000,   -1.0000000,   -1.0000000,   -1.0000000,   -1.0000000
   0.3500000,   -1.0000000,   -1.0000000,   -1.0000000,   -1.0000000
   0.5000000,   -1.0000000,   -1.0000000,   -1.0000000,   -1.0000000
   0.6500000,   -1.0000000,   -1.0000000,   -1.0000000,   -1.0000000
   0.8000000,   -1.0000000,   -1.0000000,   -1.0000000,   -1.0000000
   0.9500000,   -1.0000000,   -1.0000000,   -1.0000000,   -1.0000000
aspect_ratios
   1.0000000
   2.0000000
   0.5000000
   3.0000000
   0.3333000
In TIDL_tfliteRtImportNode  TIDL Layer type 54   Tflite builtin code type 53 
In TIDL_tfliteRtImportNode  TIDL Layer type 8   Tflite builtin code type 0 
In TIDL_tfliteRtImportNode  TIDL Layer type 8   Tflite builtin code type 18 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 5   Tflite builtin code type 0 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 5   Tflite builtin code type 0 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 5   Tflite builtin code type 0 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 5   Tflite builtin code type 0 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 5   Tflite builtin code type 0 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 5   Tflite builtin code type 0 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 5   Tflite builtin code type 0 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 5   Tflite builtin code type 0 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 5   Tflite builtin code type 0 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 5   Tflite builtin code type 0 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 5   Tflite builtin code type 0 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 5   Tflite builtin code type 0 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 5   Tflite builtin code type 0 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 5   Tflite builtin code type 0 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 5   Tflite builtin code type 0 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 4 
In TIDL_tfliteRtImportNode  TIDL Layer type 1   Tflite builtin code type 3 
In TIDL_runtimesOptimizeNet: LayerIndex = 125, dataIndex = 113 
Warning : Requested Output Data Convert Layer is not Added to the network, It is currently not Optimal
Warning : Requested Output Data Convert Layer is not Added to the network, It is currently not Optimal
Warning : Requested Output Data Convert Layer is not Added to the network, It is currently not Optimal
Warning : Requested Output Data Convert Layer is not Added to the network, It is currently not Optimal

 ************** Frame index 1 : Running float import ************* 
In TIDL_runtimesPostProcessNet 
In TIDL_runtimesPostProcessNet 1
In TIDL_runtimesPostProcessNet 2
In TIDL_runtimesPostProcessNet 3
****************************************************
**                ALL MODEL CHECK PASSED          **
****************************************************

In TIDL_runtimesPostProcessNet 4
************ in TIDL_subgraphRtCreate ************ 
 The soft limit is 2048
The hard limit is 2048
MEM: Init ... !!!
MEM: Init ... Done !!!
 0.0s:  VX_ZONE_INIT:Enabled
 0.5s:  VX_ZONE_ERROR:Enabled
 0.6s:  VX_ZONE_WARNING:Enabled
 0.1277s:  VX_ZONE_INIT:[tivxInit:185] Initialization Done !!!
Bus error (core dumped)
(3.10.16) root@cd64c3ba8cc1:/opt/edgeai-tidl-tools/examples/osrt_python/tfl#

Not sure if this is of any interest, but we base our dev-container on

nvidia/cuda:12.3-devel-ubuntu22.04

(3.10.16) root@cd64c3ba8cc1:/opt/edgeai-tidl-tools/examples/osrt_python/tfl# nvidia-smi
Tue Jan 14 14:22:21 2025       
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 555.58.02              Driver Version: 556.12         CUDA Version: 12.5     |
|-----------------------------------------+------------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
|                                         |                        |               MIG M. |
|=========================================+========================+======================|
|   0  NVIDIA RTX A2000 12GB          On  |   00000000:01:00.0  On |                  Off |
| 30%   26C    P8              5W /   70W |     255MiB /  12282MiB |      0%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+
                                                                                         
+-----------------------------------------------------------------------------------------+
| Processes:                                                                              |
|  GPU   GI   CI        PID   Type   Process name                              GPU Memory |
|        ID   ID                                                               Usage      |
|=========================================================================================|
|  No running processes found                                                             |
+-----------------------------------------------------------------------------------------+

Thanks in advance,

Dominic

0 Jianzhong Xu 16 days ago in reply to Dominic Mai

TI__Guru 50295 points

Hi Dominic,

Reese is out this week and won't be able to respond until next week.

Regards,

Jianzhong

0 Reese Grimsley 13 days ago in reply to Dominic Mai

TI__Genius 10636 points

Hi Dominic,

Thanks for the patience while I was out.

I realize that I recommended a CLI option -m that wasn't in this version of the tools. My apologies, I had forgotten this SDK's tools needed the set of models defined within the script itself in this release.

I see that you are getting the (particularly opaque) "bus error" as the script fails out. This is the case for all the models you try, correct? Generally there is an easy solution to TIDL failing on bus error. This occurs when some shared memory under /dev/shm fails to clear, and it is unable to allocate more, resulting in error. Try the line below to clear the /dev files that TIDL would have created:

rm /dev/shm/vashm*

I noted that my compilation ran into an issue in the last stage for model 'od-tfl-ssdlite_mobiledet_dsp_320x320_coco' and 'od-tfl-ssd_mobilenet_v2_300_float' (later than your logs) but the other two complete without issue and provide reasonable output. To be completely frank, 9.0 SDK was the least stable of releases between 8.6 and current (10.1) -- I recommend upgrading if possible.

I think your container is fine. Ubuntu 22.04 is correct. SDK 9.0 did not have GPU-based tools for speeding up compilation, so GPU info / status should not play a role here.

BR,
Reese

0 Dominic Mai 2 days ago in reply to Reese Grimsley

Prodigy 30 points

Hi Reese,

I tried your suggestion - unfortunately I get the same buserror one output later.

************ in TIDL_subgraphRtCreate ************ 
 The soft limit is 2048
The hard limit is 2048
MEM: Init ... !!!
MEM: Init ... Done !!!
 0.0s:  VX_ZONE_INIT:Enabled
 0.4s:  VX_ZONE_ERROR:Enabled
 0.5s:  VX_ZONE_WARNING:Enabled
 0.1520s:  VX_ZONE_INIT:[tivxInit:185] Initialization Done !!!
************ TIDL_subgraphRtCreate done ************ 
 tidl_tfLiteRtImport_delegate.cpp Invoke 478 
*******   In TIDL_subgraphRtInvoke  ******** 
Bus error (core dumped)

container resources should be good (started the container fresh, only running container)

CONTAINER ID   NAME            CPU %     MEM USAGE / LIMIT     MEM %     NET I/O        BLOCK I/O   PIDS
f21fbd32371b   loving_mclean   6.04%     1.688GiB / 31.19GiB   5.41%     86MB / 1.4MB   0B / 0B     106

Reese Grimsley said:
To be completely frank, 9.0 SDK was the least stable of releases between 8.6 and current (10.1) -- I recommend upgrading if possible.

I think this is what we'll be going to do. Stefan managed to deploy a custom trained yolox model meanwhile on version 10.

Related: I noticed that we'll likely meet on feb 10, as I'll participate in the SICK workshop where you and Manuel Philippin are signed up--> would it make sense that we compile a list of questions / topics for you beforehand?

best regards,

Dominic

0 Reese Grimsley 2 days ago in reply to Dominic Mai

TI__Genius 10636 points

Hi Dominic,

Hmm, still experiencing that bus error. I'm surprised clearing the shared memory didn't resolve this, especially if you are compiling one small model as an initial test.

I do think you'll have a much better experience in 10.0 or newer.

I will mention that a model similar to "od-tfl-ssdlite_mobiledet_dsp_320x320_coco" had an issue on 10.0/10.1. I just confirmed the fix for that last week and bugfix release with this will go live in the next week or so (10.1.0.4 is the equivalent version string for that). I mention this as a quick warning in case you see some error with "ValueError: basic_string::_M_create" prominently printed.

Dominic Mai said:
I think this is what we'll be going to do. Stefan managed to deploy a custom trained yolox model meanwhile on version 10.

Awesome, that's great to hear.

Dominic Mai said:
would it make sense that we compile a list of questions / topics for you beforehand?

Yes! That would be very helpful -- we can then get the content and discussion geared to be as practical as possible. Please send me / Manuel a list of questions so we can review and prepare. Looking forward to meeting you!

BR,
Reese

Processors

Processors forum

AM62A3: Models compiled with edgeai-tidl-tools cause segmentation fault on AM62a