/openvino

Document for Latest Intel OpenVino 2021.3 ToolKit

support: nitinashu1995@gmail.com

Base Image Creation

  • Caffe Installation

    Setup Caffe in openvino/ubuntu18_runtime:latest for the siamese model.

apt update && apt install caffe-cpu

  • Caffe Installation Verification

    After installation verify that is import Caffe working or not.

    If you are getting ImportError: cannot import name '_validate_lengths' error, degrade numpy==1.15.0 using this command

    pip3 install numpy==1.15.0

    With this change, import caffe should work. If not refer here

Model Optimizer

Use openvino/ubuntu18_dev:latest for model conversion.

  • Converting Your Caffe* Model

    cd /opt/intel/openvino_2021.3.394/deployment_tools/model_optimizer/
python3 mo_caffe.py --input_model /app/helmet_1/siamese_training_combinedhelmet_128x128__iter_200000.caffemodel 
                    --input_proto /app/helmet_1/deploy_helmet_with_dist.prototxt
                    --input_shape [1,3,128,128]
                    --output_dir /app/

Cutting parts of the model caffe model

model is too complex (contains lots of unsupported operations that cannot be easily implemented as custom layers), so the complete model cannot be converted in one shot.

single custom layer or a combination of custom layers is isolated for debugging purposes.

Model Optimizer provides command line options --input and --output to specify new entry and exit nodes, while ignoring the rest of the model:

  • --input option accepts a comma-separated list of layer names of the input model that should be treated as new entry points to the model.

  • --output option accepts a comma-separated list of layer names of the input model that should be treated as new exit points from the model.

    python3 mo_caffe.py --input_model /app/helmet_1/siamese_training_combinedhelmet_128x128__iter_200000.caffemodel
                    --input_proto /app/helmet_1/deploy_helmet_with_dist.prototxt
                     --input_shape [1,3,128,128]
                     --output_dir /app/
                     --output 'pool5/7x7_s1_right','dist'