Class Similarity Weighted Knowledge Distillation for Continual Semantic Segmentation

Deep learning models are known to suffer from the problem of catastrophic forgetting when they incrementally learn new classes. Continual learning for semantic segmentation (CSS) is an emerging field in computer vision. We identify a problem in CSS: A model tends to be confused between old and new c...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:Proceedings (IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Online) s. 16845 - 16854
Hlavní autori: Phan, Minh Hieu, Ta, The-Anh, Phung, Son Lam, Tran-Thanh, Long, Bouzerdoum, Abdesselam
Médium: Konferenčný príspevok..
Jazyk:English
Vydavateľské údaje: IEEE 01.06.2022
Predmet:
Computational modeling
Computer vision
Deep learning
grouping and shape analysis; Vision applications and systems
Machine vision
Representation learning
Scene analysis and understanding; Computer vision theory; Deep learning architectures and techniques; Efficient learning and inferences; Representation learning; Segmentation
Semantics
Shape
ISSN:1063-6919
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:Deep learning models are known to suffer from the problem of catastrophic forgetting when they incrementally learn new classes. Continual learning for semantic segmentation (CSS) is an emerging field in computer vision. We identify a problem in CSS: A model tends to be confused between old and new classes that are visually similar, which makes it forget the old ones. To address this gap, we propose REMINDER - a new CSS framework and a novel class similarity knowledge distillation (CSW-KD) method. Our CSW-KD method distills the knowledge of a previous model on old classes that are similar to the new one. This provides two main benefits: (i) selectively revising old classes that are more likely to be forgotten, and (ii) better learning new classes by relating them with the previously seen classes. Extensive experiments on Pascal-Voc 2012 and ADE20k datasets show that our approach outperforms state-of-the-art methods on standard CSS settings by up to 7.07% and 8.49%, respectively.
ISSN:1063-6919
DOI:10.1109/CVPR52688.2022.01636