Help !!! YOLOv3 !!! LeakyReLU cannot be converted to IE IR

[huunhan312]

I convert weights to pb format as following python convert_weights_pb.py --class_names coco.names --data_format NHWC --weights_file yolov3.weights --output_graph freeze_yolov3_model.pb However, the model optimizer cannot convert LeakyReLU to IR. Command: python mo_tf.py --input_model freeze_yolov3_model.pb --tensorflow_use_custom_operations_config extensions\front\tf\yolo_v3.json --input_shape [1,416,416,3] And I got the problem with LeakyRelu here

Model Optimizer arguments: Common parameters:

• Path to the Input Model: c:\Intel\computer_vision_sdk\deployment_tools\model_optimizer\freeze_yolov3_model.pb
• Path for generated IR: c:\Intel\computer_vision_sdk\deployment_tools\model_optimizer.
• IR output name: freeze_yolov3_model
• Log level: ERROR
• Batch: Not specified, inherited from the model
• Input layers: Not specified, inherited from the model
• Output layers: Not specified, inherited from the model
• Input shapes: [1,416,416,3]
• Mean values: Not specified
• Scale values: Not specified
• Scale factor: Not specified
• Precision of IR: FP32
• Enable fusing: True
• Enable grouped convolutions fusing: True
• Move mean values to preprocess section: False
• Reverse input channels: False TensorFlow specific parameters:
• Input model in text protobuf format: False
• Offload unsupported operations: False
• Path to model dump for TensorBoard: None
• List of shared libraries with TensorFlow custom layers implementation: None
• Update the configuration file with input/output node names: None
• Use configuration file used to generate the model with Object Detection API: None
• Operations to offload: None
• Patterns to offload: None
• Use the config file: c:\Intel\computer_vision_sdk\deployment_tools\model_optimizer\extensions\front\tf\yolo_v3.json Model Optimizer version: 1.5.12.49d067a0 [ ERROR ] List of operations that cannot be converted to IE IR: [ ERROR ] LeakyRelu (72) [ ERROR ] detector/darknet-53/Conv/LeakyRelu [ ERROR ] detector/darknet-53/Conv_1/LeakyRelu [ ERROR ] detector/darknet-53/Conv_2/LeakyRelu [ ERROR ] detector/darknet-53/Conv_3/LeakyRelu [ ERROR ] detector/darknet-53/Conv_4/LeakyRelu ....

System:

• Python 3.6.5
• Tensorflow 1.13.1
• OpenVINO sdk_2018.5.456

My system can run examples classification, detection of squeezenet, ssd-mobilenet well with CPU and NCS2.

Anyone please tell me any clue about this problem?

Thank you.

p/s: besides, deployment_tools\model_optimizer\extensions\front\tf only has yolo_v3.json and yolo_v1_v2.json. I cannot find yolo_v3-tiny.json as tutorial at here.

[anhminh3105]

Hi, I took reference from this repo and made my own implementation for use in combination with the NCSes, may it help you out with what you are doing. Link is here: https://github.com/anhminh3105/YOLOw

[huunhan312]

@anhminh3105 You did a great job ! Your sharing will help me a lot !