[Reopen unanswered question from July 27th - https://e2e.ti.com/support/processors-group/processors/f/processors-forum/1121579/tda4vm-tda4ve-running-into-errors-when-using-resnet50v1-in-onnx-file-format]
I wasn't able to select the TDA4VE as a part number, so I selected the TDA4VM
I'm using the software packages "ti-processor-sdk-rtos-j721s2-evm-08_00_04_04" and tried to use the resnet50 v1 network in onnx file format
During converting into the bin file following error occurs:
Conversion into bin file... ONNX Model (Proto) File : ../../test/testvecs/models/public/onnx/resnet_v1_50_224_opset11.onnx TIDL Network File : ../../test/testvecs/config/tidl_models/onnx/tidl_resnet_v1_50_224_ti_net.bin TIDL IO Info File : ../../test/testvecs/config/tidl_models/onnx/tidl_resnet_v1_50_224_ti_io_ Current ONNX OpSet Version : 11 Could not find const or initializer of layer resnet_v1_50/predictions/Reshape_1 !!! Only float and INT64 tensor is supported In put of TIDL_SoftMaxLayer layer needs to be Faltten. Please add Flatten layer to import this mdoels. Except SSD. ~~~~~Running TIDL in PC emulation mode to collect Activations range for each layer~~~~~ Processing config file #0 : /home/user/workdir/ti-processor-sdk-rtos-j721s2-evm-08_00_04_04/tidl_j721s2_08_02_00_11/ti_dl/test/testvecs/config/tidl_models/onnx/tidl_import_resnet_v1_50_224.txt.qunat_stats_config.txt Error at line: 1005 : in file /opt/Jenkins_OUT/workspace/J7_TIDL_Parent/c7x-mma-tidl/ti_dl/release/tidl_j721s2_08_02_00_11/ti_dl/test/src/pc_linux/../tidl_rt.c, of function : TIDLRT_create Error Type: TIDL_E_UNSUPPORTED_LAYER Error at line: 486 : in file /opt/Jenkins_OUT/workspace/J7_TIDL_Parent/c7x-mma-tidl/ti_dl/release/tidl_j721s2_08_02_00_11/ti_dl/test/src/pc_linux/../tidl_tb.c, of function : tidlMultiInstanceTest Error Type: TIDL_E_UNSUPPORTED_LAYER ~~~~~Running TIDL in PC emulation mode to collect Activations range for each layer~~~~~ Processing config file #0 : /home/user/workdir/ti-processor-sdk-rtos-j721s2-evm-08_00_04_04/tidl_j721s2_08_02_00_11/ti_dl/test/testvecs/config/tidl_models/onnx/tidl_import_resnet_v1_50_224.txt.qunat_stats_config.txt Error at line: 1005 : in file /opt/Jenkins_OUT/workspace/J7_TIDL_Parent/c7x-mma-tidl/ti_dl/release/tidl_j721s2_08_02_00_11/ti_dl/test/src/pc_linux/../tidl_rt.c, of function : TIDLRT_create Error Type: TIDL_E_UNSUPPORTED_LAYER Error at line: 486 : in file /opt/Jenkins_OUT/workspace/J7_TIDL_Parent/c7x-mma-tidl/ti_dl/release/tidl_j721s2_08_02_00_11/ti_dl/test/src/pc_linux/../tidl_tb.c, of function : tidlMultiInstanceTest Error Type: TIDL_E_UNSUPPORTED_LAYER ***************** Calibration iteration number 0 completed ************************ ------------------ Network Compiler Traces ----------------------------- NC running for device: 1 Running with OTF buffer optimizations: 1 Wrong data type for layer 80, 16 Wrong data type for layer 80, 16 Error : Error Code = <ERR_UNSUPPORTED_DATA_TYPE> substitute string tidl_net_ not found Could not open /home/user/workdir/ti-processor-sdk-rtos-j721s2-evm-08_00_04_04/tidl_j721s2_08_02_00_11/ti_dl/test/testvecs/config/tidl_models/onnx/tidl_resnet_v1_50_224_ti_net/perfSimInfo.bin TIDL ALLOWLISTING LAYER CHECK: [TIDL_TransposeLayer] resnet_v1_50/conv1/Conv2D__6 should be removed in import process. If not, this model will not work! TIDL ALLOWLISTING LAYER CHECK: [TIDL_SoftMaxLayer] resnet_v1_50/predictions/Softmax input shape of softmax must be 1x1x1xN. TIDL ALLOWLISTING LAYER CHECK: TIDL_E_QUANT_STATS_NOT_AVAILABLE] tidl_quant_stats_tool.out fails to collect dynamic range. Please look into quant stats log. This model will get fault on target. WARNING: [TIDL_E_DATAFLOW_INFO_NULL] ti_cnnperfsim.out fails to allocate memory in MSMC. Please look into perfsim log. This model can only be used on PC emulation, it will get fault on target. **************************************************** ** 1 WARNINGS 3 ERRORS ** ****************************************************
I wasn't able to upload the onnx file, maybe because of the file size of about 100MByte. Please let me know if I can upload somewhere if the file is needed.
Do you have any idea how I can overcome this issue without changing the network?
I already tried it with onnx opset 11 and 13
Thanks,
Martin
