I'm trying to get the demo running from here...
But somehow the pre-trained model from Intel won't run. I'm on Mac OSX and 2021.1 now.
rloÉrlo-mac01 application % python3 ./safety_gear_detector.py -m ../intel/person-detection-retail-0013/FP16/person-detection-retail-0013.xml Application running in Async mode /Users/rlo/Documents/python_demos/safety-gear-detector-python/application/inference.py:79: DeprecationWarning: 'layers' property of IENetwork class is deprecated. For iteration over network please use get_ops()/get_ordered_ops() methods from nGraph Python API Äl for l in self.net.layers.keys() if l not in supported_layersÅ ERROR:root:Following layers are not supported by the plugin for specified device CPU: Constant_58577
I actually can run this on the neural stick... Same code won't run if I switch to -d CPU....
python3 ./safety_gear_detector.py -d MYRIAD -m /opt/intel/openvino/deployment_tools/open_model_zoo/tools/downloader/intel/person-detection-retail-0013/FP16/person-detection-retail-0013.xml -sm ../resources/worker-safety-mobilenet/FP16/worker_safety_mobilenet.xml Application running in Async mode /Users/rlo/Documents/python_demos/safety-gear-detector-python/application/inference.py:97: DeprecationWarning: 'inputs' property of IENetwork class is deprecated. To access DataPtrs user need to use 'input_data' property of InputInfoPtr objects which can be accessed by 'input_info' property. self.input_blob = next(iter(self.net.inputs)) /Users/rlo/Documents/python_demos/safety-gear-detector-python/application/inference.py:153: DeprecationWarning: 'outputs' property of InferRequest is deprecated. Please instead use 'output_blobs' property.
I suggest you run the sample code with OpenVINO 2020.3 since this has been mentioned requirement of the sample. Meanwhile, make sure your CPU is within 6th to 8th Generation Intel® Core™ processors with Iris® Pro graphics or Intel® HD Graphics which also mentioned in the hardware requirement. The sample might be hardcoded to OpenVINO 2020.3 and 6th to 8th generation core CPU.
That's a huge uptake to roll back all versions and my setup back to 2019 to get it running.
I don't see where it did the hardcoding either. It actually compiles and optimizes the model with the mo_caffe.py script.
cd /opt/intel/openvino/deployment_tools/model_optimizer/ ./mo_caffe.py --input_model $BASE_DIR/resources/worker-safety-mobilenet/worker_safety_mobilenet.caffemodel -o $BASE_DIR/resources/worker-safety-mobilenet/FP32 --data_type FP32 ./mo_caffe.py --input_model $BASE_DIR/resources/worker-safety-mobilenet/worker_safety_mobilenet.caffemodel -o $BASE_DIR/resources/worker-safety-mobilenet/FP16 --data_type FP16
I got the same error rolling back to 2020.3 LTS. Something isn't adding up. I guess I'm missing some sort of cpu_extension as I saw similar issues with others.
I will give it a try again and update you all with another machine.
This sample is a specific sample and it is not part of the OpenVINO toolkit but the intel-iot-devkit. However, I've tried in a different machine and different version of OpenVINO but it only works with OpenVINO 2020.3. If you getting any error from OpenVINO 2020.3 too, please try the workaround first.