I get problem when i use mo_tf.py to covert the my training model (based on ssd_mobilenet_v1_coco_2017_11_17). I try to use --tensorflow_object_detection_api_pipeline_config option to load my pipeline.config file, the error happened: "error: unrecognized arguments. "
Model Optimizer version: 184.108.40.206f62983
Shall I update the version?
Could you please give more details like the command line you are using to convert your model, the model files, and the version of OpenVINO that you are using?
Thank you for your reply.
I resolved the problem by update my "OpenVINO™ toolkit from 2018 R2 --> 2018 R3".
It seems that Model Optimizer in 2018 R2 not support the parameter "tensorflow_object_detection_api_pipeline_config" but R3 support. I can successfully convert the tensor flow mobilenet to IR format.
But I get a problem that the performance of NCS is very low, I will open another thread, hope you can help to answer!