search

FishYuLi / BalancedGroupSoftmax

CVPR 2020 oral paper: Overcoming Classifier Imbalance for Long-tail Object Detection with Balanced Group Softmax.
Apache License 2.0
351 stars 64 forks source link

How to train my own dataset? #5

Open CV-deeplearning opened 4 years ago

CV-deeplearning commented 4 years ago

Thank you for your great project. But how to train my own dataset?

FishYuLi commented 4 years ago

Hi, @CV-deeplearning. If you want to train your own dataset, you need to:

  1. Generate BAGS models: label2binlabel.pt, pred_slice_with0.pt, valsplit.pkl these 3 intermediate files referring to these 2 functions (1, 2).
  2. Let your config file point to your generated files.
  3. Modify the category number.

Then you can train on your own dataset.