This script allows you to reshape the OpenVINO model input layer.
This way, if your model normally accepts 300x300 input, you can change the dimensions to any value, e.x. 200x200
docker build -t reshaper .
docker run -v $(pwd)/model:/home/openvino/model -e RESHAPE=200x200 -e MODEL_PATH=/home/openvino/model/face-detection-retail-0004.xml -e WEIGHTS_PATH=/home/openvino/model/face-detection-retail-0004.bin --rm reshaper
reshaped files should be available inside model directory
source /opt/intel/openvino/bin/setupvars.sh # activate OpenVINO
python3 reshape_openvino_model.py -m $(pwd)/model/face-detection-retail-0004.xml -w $(pwd)/model/face-detection-retail-0004.bin -r 200x200