search

schuy1er / EWF_official

An official code for "Endpoints Weight Fusion for Class Incremental Semantic Segmentation"
MIT License
25 stars 4 forks source link

Fixed pre_cls_weight #3

Closed jainie-max closed 1 year ago

jainie-max commented 1 year ago

Hi！ Thanks for your great work, I have a small question about the fixed pre_cls classifier weight. Why the weight of pre_cls in the classifier need fixed during the current step?

schuy1er commented 1 year ago

Hi! Thanks for your question. The main reason is that fixing it will enhance the stability of the method. This is equivalent to adjusting the learning rate on the classifiers. I mentioned this step in the implementation details in the supplementary material.