Skip to content

Latest commit

 

History

History
26 lines (9 loc) · 1.34 KB

File metadata and controls

26 lines (9 loc) · 1.34 KB

Pipeline for detecting and analyzing body parts of black and white mice using mask RCNN and inception coco. Can be used for 'fingerprinting' animals and individual identification.

alt-text-1 alt-text-1

  • Weights for mouse resident intruder can be downloaded on OSF.
  • Uses a modified version of labelme2coco to create coco.json formats from labelme output.
  • Uses a modified version of create_coco_tf_record.py to create tf records from coco.json format.
  • Example renderings on youtube: [1], [2], [3].
  • Weights were created by labelling images using labelme and labelling 9 body-parts on each mouse. alt-text-1