An incremental object detection model based on knowledge distillation

Huimin Liao; Jingming Luo; Jinghui Zhang

doi:10.1117/12.3009226

11 October 2023 An incremental object detection model based on knowledge distillation

Huimin Liao, Jingming Luo, Jinghui Zhang

Proceedings Volume 12918, Fourth International Conference on Computer Science and Communication Technology (ICCSCT 2023); 129180I (2023) https://doi.org/10.1117/12.3009226
Event: International Conference on Computer Science and Communication Technology (ICCSCT 2023), 2023, Wuhan, China

Abstract

This paper proposes an incremental object detection model framework based on knowledge distillation. The framework addresses the issue of catastrophic forgetting by optimizing the distillation loss function and expanding the labels of the new dataset based on the old object classes, which significantly enhances the recognition capability of the model for the old objects. The framework is implemented and evaluated on the two-stage object detection algorithm. Experimental results on the PASCAL VOC dataset demonstrate that the proposed model achieves competitive performance compared to state-of-the-art methods.

(2023) Published by SPIE. Downloading of the abstract is permitted for personal use only.

Citation Download Citation

Huimin Liao, Jingming Luo, and Jinghui Zhang "An incremental object detection model based on knowledge distillation", Proc. SPIE 12918, Fourth International Conference on Computer Science and Communication Technology (ICCSCT 2023), 129180I (11 October 2023); https://doi.org/10.1117/12.3009226

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available