Skip to content

Latest commit

 

History

History
16 lines (15 loc) · 816 Bytes

README.md

File metadata and controls

16 lines (15 loc) · 816 Bytes

eetfm_automation

eetfm_automation: Export and Evaluate TensorFlow Model Automation based on TensorFlow Object Detection API.

Note: Currently only faster_rcnn based models supported (e.g mask_rcnn)

How to use this repository

  • Place standard TF Object_detection API model directories inside the base_models folder, containing at least the following files:
    model.ckpt.data-00000-of-00001
    model.ckpt.index
    model.ckpt.meta
    pipeline.config
    
  • Create a copy of config.sample.yml named config.yml and modify the params to your needs
  • run python eetfm_automation.py
  • the new models are exported to the export_models folder
  • evaluation results are saved to a new eval folder inside each model