OpenVINO Serving Container(NCS) - "Cannot find plugin to use" for YOLO V3 model

[Carlyon__Shane]

I converted the darknet YOLO V3 model into TF model using this - https://github.com/mystic123/tensorflow-yolo-v3.git  and optimized by this command:

python3 /opt/intel/openvino/deployment_tools/model_optimizer/mo_tf.py --input_model frozen_darknet_yolov3_model.pb --tensorflow_use_custom_operations_config /opt/intel/openvino/deployment_tools/model_optimizer/extensions/front/tf/yolo_v3.json --input_shape=[1,416,416,3] --data_type=FP16

Then I tried to serve this model with OpenVINO Serving Container(with NCS) by this command:

docker run --rm -d -it --net=host --privileged -v "$(pwd)"/models/:/opt/model:z -v /dev:/dev -e DEVICE=MYRIAD -p 9001:9001 intelaipg/openvino-model-server:latest /ie-serving-py/start_server.sh ie_serving model --model_path /opt/model/darknet_yolov3 --model_name darknet_yolov3 --port 9001

But I am getting this error from docker log:

2019-06-09 08:51:15,732 - ie_serving.models.model - INFO - Server start updating model: darknet_yolo_v3_ncs
2019-06-09 08:51:15,732 - ie_serving.models.model - INFO - Creating inference engine object for version: 1
2019-06-09 08:51:15,732 - ie_serving.models.model - ERROR - Error occurred while loading model version: {'xml_file': '/opt/model/darknet_yolo_v3_ncs/1/frozen_darknet_yolov3_model.xml', 'bin_file': '/opt/model/darknet_yolo_v3_ncs/1/frozen_darknet_yolov3_model.bin', 'mapping_config': None, 'version_number': 1, 'batch_size': None}
2019-06-09 08:51:15,732 - ie_serving.models.model - ERROR - Content error: Cannot find plugin to use :
2019-06-09 08:51:15,733 - ie_serving.models.model - INFO - List of available versions after updatefor darknet_yolo_v3_ncs model: []
2019-06-09 08:51:15,733 - ie_serving.models.model - INFO - Default version after updatefor darknet_yolo_v3_ncs model is -1

Any idea?

[Carlyon__Shane]

UPDATED:

I even tried this link: https://software.intel.com/en-us/forums/computer-vision/topic/805370#comment-1934835

But I am getting this error this time:

work@work-MS-7788:~$ docker run --rm -d -it -v /home/work/Downloads/models:/opt/ml/:ro -p 9001:9001 intelaipg/openvino-model-server:latest /ie-serving-py/start_server.sh ie_serving model --model_path /opt/ml/darknet_yolov3 --model_name darknet_yolov3 --port 9001
d9b907250fe457fc8096f12913f39e71f8dfcebcc965237891986bed01fab841
work@work-MS-7788:~$ docker logs d9
2019-06-10 07:48:15,038 - ie_serving.main - INFO - Log level set: INFO
2019-06-10 07:48:15,038 - ie_serving.models.model - INFO - Server start loading model: darknet_yolov3
2019-06-10 07:48:15,040 - ie_serving.models.model - INFO - Creating inference engine object for version: 1
2019-06-10 07:48:15,053 - ie_serving.models.model - ERROR - Error occurred while loading model version: {'xml_file': '/opt/ml/darknet_yolov3/1/frozen_darknet_yolov3_model.xml', 'bin_file': '/opt/ml/darknet_yolov3/1/frozen_darknet_yolov3_model.bin', 'mapping_config': None, 'version_number': 1, 'batch_size': None}
2019-06-10 07:48:15,053 - ie_serving.models.model - ERROR - Content error: Error reading network: in Layer yolov3/convolutional2/Conv2D: trying to connect an edge to non existing output port: 2.1
2019-06-10 07:48:15,054 - ie_serving.models.model - INFO - List of available versions for darknet_yolov3 model: []
2019-06-10 07:48:15,054 - ie_serving.models.model - INFO - Default version for darknet_yolov3 model is -1
2019-06-10 07:48:15,060 - ie_serving.server.start - INFO - Server listens on port 9001 and will be serving models: ['darknet_yolov3']

Any idea?

[Shubha_R_Intel]

Dear Carlyon, Shane,

The latest OpenVino release is 2019R1.1. Yolo v3 definitely works on it, though Tiny Yolo V3 is broken. Did you follow Tensorflow MO Yolo V3 steps ? Please try again with the latest and greatest OpenVino. I recently tried myself and I know that Yolo V3 (non-Tiny) works.

Glad that you resolved your "Can't find Plugin" problem. That is always related to your plugin binary directory not being in your $PATH variable.

Thanks,

Shubha