model converter

OpenVINOで提供されているモデルでは、Inference Engine IR Formatが提供されていないものもあります

IR Formatのモデルは、Open Model Zoo内のサンプルでも使用します

model converterは、

 ./ -h
 usage: [-h] [-c CONFIG.YML] [-d DIR] [-o DIR]
                     [--name PAT[,PAT...]] [--list FILE.LST] [--all]
                     [--print_all] [--precisions PREC[,PREC...]] [-p PYTHON]
                     [--mo MO.PY] [--add-mo-arg ARG] [--dry-run] [-j JOBS]

 optional arguments:
   -h, --help            show this help message and exit
   -c CONFIG.YML, --config CONFIG.YML
                         model configuration file (deprecated)
   -d DIR, --download_dir DIR
                         root of the directory tree with downloaded model files
   -o DIR, --output_dir DIR
                         root of the directory tree to place converted files
   --name PAT[,PAT...]   convert only models whose names match at least one of
                         the specified patterns
   --list FILE.LST       convert only models whose names match at least one of
                         the patterns in the specified file
   --all                 convert all available models
   --print_all           print all available models
   --precisions PREC[,PREC...]
                         run only conversions that produce models with the
                         specified precisions
   -p PYTHON, --python PYTHON
                         Python executable to run Model Optimizer with
   --mo MO.PY            Model Optimizer entry point script
   --add-mo-arg ARG      Extra argument to pass to Model Optimizer
   --dry-run             Print the conversion commands without running them
   -j JOBS, --jobs JOBS  number of conversions to run concurrently 


/opt/intel/openvino/deployment_tools/open_model_zoo/tools/downloader/ --name ssd300 -d ~/models/ --precisions=FP16

========= Converting ssd300 to IR (FP16)  Conversion command: /usr/bin/python3 -- /opt/intel/openvino_2020.1.023/deployment_tools/model_optimizer/ --framework=caffe --data_type=FP16 --output_dir=/home/klf/models/public/ssd300/FP16 --model_name=ssd300 '--input_shape=[1,3,300,300]' --input=data '--mean_values=data[104.0,117.0,123.0]' --output=detection_out --input_model=/home/klf/models/public/ssd300/models/VGGNet/VOC0712Plus/SSD_300x300_ft/VGG_VOC0712Plus_SSD_300x300_ft_iter_160000.caffemodel --input_proto=/home/klf/models/public/ssd300/models/VGGNet/VOC0712Plus/SSD_300x300_ft/deploy.prototxt    Model Optimizer arguments:  Common parameters:  - Path to the Input Model:  /home/klf/models/public/ssd300/models/VGGNet/VOC0712Plus/SSD_300x300_ft/VGG_VOC0712Plus_SSD_300x300_ft_iter_160000.caffemodel  
- Path for generated IR:  /home/klf/models/public/ssd300/FP16  
- IR output name:  ssd300  
- Log level:  ERROR  
- Batch:  Not specified, inherited from the model  
- Input layers:  data  - Output layers:  detection_out  
- Input shapes:  [1,3,300,300]  
- Mean values:  data[104.0,117.0,123.0]  
- Scale values:  Not specified  
- Scale factor:  Not specified  
- Precision of IR:  FP16  
- Enable fusing:  True  
- Enable grouped convolutions fusing:  True  
- Move mean values to preprocess section:  False  
- Reverse input channels:  False  Caffe specific parameters:  
- Path to Python Caffe* parser generated from caffe.proto:  /opt/intel/openvino_2020.1.023/deployment_tools/model_optimizer/mo/front/caffe/proto  
- Enable resnet optimization:  True  
- Path to the Input prototxt:  /home/klf/models/public/ssd300/models/VGGNet/VOC0712Plus/SSD_300x300_ft/deploy.prototxt  - Path to CustomLayersMapping.xml:  Default  - Path to a mean file:  Not specified  - Offsets for a mean file:  Not specified  Model Optimizer version:  2020.1.0-61-gd349c3ba4a    
[ SUCCESS ] Generated IR version 10 model.  
[ SUCCESS ] XML file: /home/klf/models/public/ssd300/FP16/ssd300.xml  
[ SUCCESS ] BIN file: /home/klf/models/public/ssd300/FP16/ssd300.bin  
[ SUCCESS ] Total execution time: 11.24 seconds.   
[ SUCCESS ] Memory consumed: 585 MB.