I'd like to optimise a yolov3 model, made using darknet.
But following this tuto (https://software.intel.com/en-us/articles/OpenVINO-Using-TensorFlow#converting-a-darknet-yolo-model) to optimise my darknet model, I got some ERRORS:
# sudo python3 mo_tf.py --input_model /home/quentin/ConvertYOLOv3TOIR/tensorflow-yolo-v3/frozen_darknet_yolov3_model.pb --tensorflow_use_custom_operations_config /home/quentin/ConvertYOLOv3TOIR/OpenVino/computer_vision_sdk_2018.5.455/deployment_tools/model_optimizer/extensions/front/tf/yolo_v3.json
Model Optimizer version: 220.127.116.11d067a0
[ ERROR ] Shape [ -1 416 416 3] is not fully defined for output 0 of "inputs". Use --input_shape with positive integers to override model input shapes.
[ ERROR ] Cannot infer shapes or values for node "inputs".
[ ERROR ] Not all output shapes were inferred or fully defined for node "inputs".
For more information please refer to Model Optimizer FAQ (<INSTALL_DIR>/deployment_tools/documentation/docs/MO_FAQ.html), question #40.
[ ERROR ]
[ ERROR ] It can happen due to bug in custom shape infer function <function tf_placeholder_ext.<locals>.<lambda> at 0x7fa19f7e48c8>.
[ ERROR ] Or because the node inputs have incorrect values/shapes.
[ ERROR ] Or because input shapes are incorrect (embedded to the model or passed via --input_shape).
[ ERROR ] Run Model Optimizer with --log_level=DEBUG for more information.
[ ERROR ] Stopped shape/value propagation at "inputs" node.
For more information please refer to Model Optimizer FAQ (<INSTALL_DIR>/deployment_tools/documentation/docs/MO_FAQ.html), question #38.
I tried the tutorial with the same model as advised: yolov3.weights, coco.names but I got the same errors.
Any help would be appreciated!
Thanks and best regards!
The error tells you exactly what the problem is:
[ ERROR ] Shape [ -1 416 416 3] is not fully defined for output 0 of "inputs". Use --input_shape with positive integers to override
Either pass in something like --input_shape [1, 416, 416, 3] or pass in --batch 1, where 1 is batch-size. I am using 1 as batch-size but you can choose to use another positive number. That -1 in the batch-size position is the issue which Model Optimizer is complaining about.
Thanks for using OpenVino !