Skip to content

Latest commit

 

History

History
57 lines (32 loc) · 2.04 KB

File metadata and controls

57 lines (32 loc) · 2.04 KB

Version History & Work Progress

This gives an overview of the evolution of the training process, most of those experiments didn't not give any results but they demonstrate how I progressed through this project work, until I got the final trained model in experiment pig_v4.

pig5_21

  • duroc_6stk db, 21 images annotated, no background, no masks

pig5_v1

  • duroc_6stk db, 21 images annotated, with background, no masks

pig5_v2.1

  • duroc_3stk db, 22 images annotated, with background, with masks (maybe offset by 1)

pig5_v2.2

  • duroc_3stk db, 55 images annotated, with background, with masks (mistake -> masks are offset by 1)

pig5_v2.3

  • duroc_3stk db, 55 images annotated, with background, with masks

pig5_v2.4

  • same as v2.3 but fixed visibility and mask indexes

pig5_v2.5

  • Small tweaks done after v2.4, no real improvement over the result

pig5_v2.5.1

  • Same as v2.5, training resumed from checkpoint at 6k iterations

pig5_v3

  • now with 236% more images annotated! (55 -> 130)
  • Notes: Training loss diverged after only 10k iterations

pig5_v4 (best & final)

  • another test without background and no scaling (as apparent size of the pigs does not change in the pen)
  • Trained model download link: pose_iter_124000.caffemodel

Further work & Improvement ideas

  • Images sequences are not ideal for traning because successive frames might be too similar from each other
  • The nose is rarely visible, maybe try a new model (PIG_4) with only: LEar, REar, Neck and Tail? Or add other useful joints (shoulders, hips,...)
  • Combine training data from multiples folders (3stk,6stk,10stk... duroc/landsvin...) for better generalization
  • Tweak training hyperparameters

Validation/Testing Metrics

  • Average Precision (& Average Recall), using Keypoint Similarity Score OKS from COCO
  • Proportion of # tracked entity against the ground thruth # of entities