The authors propose Restoring without Forgetting (RwF), a continual learning framework for image restoration.
Image restoration can learn new tasks without forgetting by updating only critical filters
The authors' RwF framework localizes task-specific filters with parameter-space integrated gradients and generates them via low-rank transformations, matching all-in-one quality on six restoration tasks.
Chinese Tech
Xin Feng · Jin Zhao · Yizhen Zhang · Wenjie Pei · Fanglin Chen · Guangming Lu
University of Edinburgh · Baidu Inc. · Tsinghua University · Harbin Institute of Technology, Shenzhen
Research Digest··3 min read
The authors propose Restoring without Forgetting (RwF), a continual learning method for image restoration that adapts models to sequential degradation tasks without accessing past data.
Why this paper
From Baidu Inc. and 3 others
In one line
A filter-level continual learning framework avoids forgetting in image restoration by localizing and updating only degradation-critical filters.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ·No stated limitations found
- ·No benchmark numbers found
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§