When I run mask_rcnn_inception_v2 on Myriad, the result is strange.
It happens with OpenVINO2020.2.
OpenVINO2020.1 gives correct results.(The result is the same as CPU or GPU.)
Where to get the model
I hope it will be fixed in the next version.
Thanks for reaching out.
Could you please tell us which sample/demo are you using? Also, provide us with the Model Optimizer command you used to convert the frozen model to IR format.
mask_rcnn_demo.exe -d MYRIAD -m mask_rcnn_inception_v2.xml -i test.jpg
■Model Optimizer command
--input_model frozen_inference_graph.pb" ^
--tensorflow_object_detection_api_pipeline_config pipeline.config ^
--data_type FP16 ^
--batch 1 ^
--tensorflow_use_custom_operations_config ".\extensions\front\tf\mask_rcnn_support.json" ^
Thank you for your reply.
This demo is not supported on VPU devices, that is why it may or may not work. You can check the Readme.md in the demo directory here: mask_rcnn_demo.