Hi-FT / ERD

Overcoming Catastrophic Forgetting in Incremental Object Detection via Elastic Response Distillation

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Will new-class data be feed in the teacher model?

enmengyi opened this issue · comments

  1. In a traning phase, will both old-class data and new-class data be feed in the teacher model?
  2. How do you sample the old-class data, or all the old-class data will be uesd?