TDA2PHV-ACD: TF Model Import Error

Hello Kumar,

  I tried to convert a Tensorflow slim model(Mobilenet V2) to TIDL format.The TF model I am trying to covert can be downloaded from the following path:

https://storage.googleapis.com/mobilenet_v2/checkpoints/mobilenet_v2_1.4_224.tgz

The above directory contains mobilenet_v2_1.4_224_frozen.pb which I further optimize by the following command:

python optimize_for_inference.py --input=D:\ADAS_Team\

models\mobilenet_v2_1.4_224\mobilenet_v2_1.4_224\mobilenet_v2_1.4_224_frozen.pb

--output=mobilenet_frozen_optimized_v2_1.4_224.pb --input_names=input -

-output_name=MobilenetV2/Predictions/Reshape_1

After executing the above command I get the optiimized frozen file mobilenet_frozen_optimized_v2_1.4_224.pb.

I have used this file as an input for the TIDL model import tool(tidl_model_import.out.exe) to generate the parameter and network files.I have used the attached config file for TIDL translation.

# Default - 0
randParams         = 0 

# 0: Caffe, 1: TensorFlow, Default - 0
modelType          = 1

# 0: Fixed quantization By tarininng Framework, 1: Dyanamic quantization by TIDL, Default - 1
quantizationStyle  = 1 

# quantRoundAdd/100 will be added while rounding to integer, Default - 50
quantRoundAdd      = 50

numParamBits       = 8
 
inputNetFile       = "..\..\test\testvecs\config\tensorflow_models\mobilenet\mobilenet_frozen_optimized_v2_1.4_224.pb"
inputParamsFile    = "NA"
outputNetFile      = "..\..\test\testvecs\config\tidl_models\tidl_net_mobilenet_v2_1.0_224_optimized.bin"
outputParamsFile   = "..\..\test\testvecs\config\tidl_models\tidl_param_mobilenet_v2_1.0_224_optimized.bin"

preProcType  = 2
sampleInData = "..\..\test\testvecs\input\airshow.jpg"
tidlStatsTool = "..\quantStatsTool\eve_test_dl_algo.out.exe"

inWidth  = 600
inHeight = 600 
inNumChannels = 3
conv2dKernelType = 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0

After running the TIDL model import tool with the attached config file,I get the following error:

Could not find the requested input Data : MobilenetV2/Conv/Relu6 !!

I request you to assist me in solving the error.

Best Regards,

Vijay

Hello Kumar,
I was not able to revert back to TF version 1.1.When I tried to install TF 1.1 I got the following error:
C:\Users\invkrish02>pip install tensorflow==1.1
Collecting tensorflow==1.1
Could not find a version that satisfies the requirement tensorflow==1.1 (from
versions: 1.2.0rc2, 1.2.0, 1.2.1, 1.3.0rc0, 1.3.0rc1, 1.3.0rc2, 1.3.0, 1.4.0rc0,
1.4.0rc1, 1.4.0, 1.5.0rc0, 1.5.0rc1, 1.5.0, 1.5.1, 1.6.0rc0, 1.6.0rc1, 1.6.0, 1
.7.0rc0, 1.7.0rc1, 1.7.0, 1.7.1, 1.8.0rc0, 1.8.0rc1, 1.8.0, 1.9.0rc0, 1.9.0rc1,
1.9.0rc2, 1.9.0, 1.10.0rc0, 1.10.0rc1, 1.10.0, 1.11.0rc0, 1.11.0rc1, 1.11.0rc2,
1.11.0, 1.12.0rc0, 1.12.0rc1, 1.12.0rc2, 1.12.0, 1.12.2, 1.13.0rc0, 1.13.0rc1, 1
.13.0rc2, 1.13.1, 2.0.0a0)
No matching distribution found for tensorflow==1.1

I guess version 1.1 is no longer available.Does the TIDL module work with higher versions of TF?

I tried freezing the model using TF 1.11 and I got the following errors:

Loaded meta graph file 'D:\ADAS_Team\models\mobilenet_v2_1.4_224\mobilenet_v2_1.
4_224\mobilenet_v2_1.4_224.ckpt.meta
2019-04-23 14:04:50.330434: I tensorflow/core/platform/cpu_feature_guard.cc:141]
Your CPU supports instructions that this TensorFlow binary was not compiled to
use: AVX AVX2
2019-04-23 14:04:50.425443: I tensorflow/core/common_runtime/process_util.cc:69]
Creating new thread pool with default inter op setting: 8. Tune using inter_op_
parallelism_threads for best performance.
Traceback (most recent call last):
File "freeze_graph.py", line 492, in <module>
run_main()
File "freeze_graph.py", line 489, in run_main
app.run(main=my_main, argv=[sys.argv[0]] + unparsed)
File "C:\Users\invkrish02\AppData\Local\Continuum\anaconda3\envs\tf1.1_env\lib
\site-packages\tensorflow\python\platform\app.py", line 125, in run
_sys.exit(main(argv))
File "freeze_graph.py", line 488, in <lambda>
my_main = lambda unused_args: main(unused_args, flags)
File "freeze_graph.py", line 382, in main
flags.saved_model_tags, checkpoint_version)
File "freeze_graph.py", line 364, in freeze_graph
checkpoint_version=checkpoint_version)
File "freeze_graph.py", line 155, in freeze_graph_with_def_protos
input_meta_graph_def, clear_devices=True)
File "C:\Users\invkrish02\AppData\Local\Continuum\anaconda3\envs\tf1.1_env\lib
\site-packages\tensorflow\python\training\saver.py", line 1939, in import_meta_g
raph
**kwargs)
File "C:\Users\invkrish02\AppData\Local\Continuum\anaconda3\envs\tf1.1_env\lib
\site-packages\tensorflow\python\framework\meta_graph.py", line 744, in import_s
coped_meta_graph
producer_op_list=producer_op_list)
File "C:\Users\invkrish02\AppData\Local\Continuum\anaconda3\envs\tf1.1_env\lib
\site-packages\tensorflow\python\util\deprecation.py", line 454, in new_func
return func(*args, **kwargs)
File "C:\Users\invkrish02\AppData\Local\Continuum\anaconda3\envs\tf1.1_env\lib
\site-packages\tensorflow\python\framework\importer.py", line 391, in import_gra
ph_def
_RemoveDefaultAttrs(op_dict, producer_op_list, graph_def)
File "C:\Users\invkrish02\AppData\Local\Continuum\anaconda3\envs\tf1.1_env\lib
\site-packages\tensorflow\python\framework\importer.py", line 158, in _RemoveDef
aultAttrs
op_def = op_dict[node.op]
KeyError: 'InfeedEnqueueTuple'

Best Regards,

Vijay