The Attempt on Combining Three Talents by KD with Enhanced Boundary in Co-Salient Object Detection


Ziyi Cao (Nanjing University of Information Science and Technology), Shengye Yan (Nanjing University of Information Science and Technology), Wei Zheng (MINIEYE)
The 35th British Machine Vision Conference
PDFPoster

Abstract

In the field of Co-Salient Object Detection (CoSOD), we have identified three methods that perform best on three respective test sets: GCoNet+, DMT and MCCL. Therefore, by using knowledge distillation (KD), we employ two of these methods as teacher models and one as the student model, amalgamating their knowledge to create a more powerful model. Additionally, we propose a dual-branch learning architecture for the student model, where one branch learns the consensus features of objects as a whole, and the other branch learns the consensus features of object boundaries. Experimental results demonstrate that our network achieves, or even surpasses, the current state-of-the-art performance on three widely used CoSOD benchmark datasets.

Citation

@inproceedings{Cao_2024_BMVC,
author    = {Ziyi Cao and Shengye Yan and Wei Zheng},
title     = {The Attempt on Combining Three Talents by KD with Enhanced Boundary in Co-Salient Object Detection},
booktitle = {35th British Machine Vision Conference 2024, {BMVC} 2024, Glasgow, UK, November 25-28, 2024},
publisher = {BMVA},
year      = {2024},
url       = {https://papers.bmvc2024.org/0622.pdf}
}


Copyright © 2024 The British Machine Vision Association and Society for Pattern Recognition
The British Machine Vision Conference is organised by The British Machine Vision Association and Society for Pattern Recognition. The Association is a Company limited by guarantee, No.2543446, and a non-profit-making body, registered in England and Wales as Charity No.1002307 (Registered Office: Dept. of Computer Science, Durham University, South Road, Durham, DH1 3LE, UK).

Imprint | Data Protection