Problem working with faster_rcnn_resnet50 using movidius

[Stankov__Stanko]

Hello,

So i have been struggling with this for a while and i decided to ask here.

When i try to run object detection using faster_rcnn_resnet50 i get the following error - 

Traceback (most recent call last):
  File "openvino_real_time_object_detection.py", line 74, in <module>
    detections = net.forward()
cv2.error: OpenCV(4.1.0-openvino) /home/jenkins/workspace/OpenCV/OpenVINO/build/opencv/modules/dnn/src/dnn.cpp:2299: error: (-215:Assertion failed) inp.total() in function 'allocateLayers'

I successfully get the xml and bin file using the following command

python mo_tf.py --input_model=D:/Software_development_tools/tensorflow/models-master/research/object_detection/inference_graph/frozen_inference_graph.pb --tensorflow_use_custom_operations_config C:/IntelSWTools/openvino_2019.1.133/deployment_tools/model_optimizer/extensions/front/tf/faster_rcnn_support_api_v1.13.json  --tensorflow_object_detection_api_pipeline_config D:/Software_development_tools/tensorflow/models-master/research/object_detection/inference_graph/pipeline.config --reverse_input_channels --data_type=FP16

In additional i have tried using the xml and bin files on the openvino_fd_myriad.py code provided on this page - https://docs.openvinotoolkit.org/latest/_docs_install_guides_installing_openvino_raspbian.html and i still get the same result.

I have attached the config file, the xml and bin as well as the source code.

Thank you for your time !

Best regards,

Stanko Stankov

[Shubha_R_Intel]

Dear Stankov, Stanko

This is not a traditional OpenVino problem which is concerned with only Inference Engine or Model Optimizer. But I googled your OpenCV issue and found the following (it looks exactly like your error).

https://github.com/opencv/opencv/issues/13751

OpenCV github is the best place to post your OpenCV issues.

Also there is OpenCV Questions .

Thanks !

Shubha

[Stankov__Stanko]

Dear Shubha,

Thank you for your reply!

I will do some more research and write there as well and let you know the results.

Edit : I would also like to point out that running the code with the following configuration i get a different error

net = cv2.dnn.readNetFromTensorflow('frozen_inference_graph.pb','frozen_inference_graph.pbtxt')
net.setPreferableTarget(cv2.dnn.DNN_TARGET_MYRIAD)

message : 

terminate called after throwing an instance of 'InferenceEngine::details::InferenceEngineException'
  what():  ConfidenceThreshold parameter is wrong in layer detection_out. It should be > 0.
Aborted
"
terminate called after throwing an instance of 'InferenceEngine::details::InferenceEngineException'
  what():  ConfidenceThreshold parameter is wrong in layer detection_out. It should be > 0.
Aborted

"

Best regards,

Stanko Stankov

[Shubha_R_Intel]

Dear Stankov, Stanko

Thanks for sharing back to the community. Perhaps one of those OpenCV forums will set you straight.

Thanks for your patience !

Shubha

[Stankov__Stanko]

Dear Shubha,

I have opened a thread about this in the OpenCV community and this issue is already labeled as a bug.

In case someone else has the same issue this is the link to the thread - https://github.com/opencv/opencv/issues/14839

Thank you for your time!

Best regards,

Stanko Stankov

[Shubha_R_Intel]

Dearest Stanko Stankov,

Wow. I really appreciate your sharing your findings back to the community. It's very thoughtful of you !

Thanks again for using OpenVino,

Shubha

[Sapiain__Roberto]

Hi.

Thank you for this thread. I'm having the same issue with Openvino 2019.R2,

Neural-net: Faster-RCNN topology, trained TFOD-API v1.12, with Tensorflow 1.12.3 (CUDA 9.0 patch level 4, cuDNN v7.5). 3 categories only.

Hope there is a fix soon.