Efficient vocal cord lesion recognition by combing Yolov7 and attention module

Yanda Wu; Yuqing He; Dongyan Huang; Yang Liu; Jingxuan Zhu; Hengli Zhang

doi:10.1117/12.3007662

18 December 2023 Efficient vocal cord lesion recognition by combing Yolov7 and attention module

Yanda Wu, Yuqing He, Dongyan Huang, Yang Liu, Jingxuan Zhu, Hengli Zhang

Proceedings Volume 12963, AOPC 2023: Optical Sensing, Imaging, and Display Technology and Applications; and Biomedical Optics; 129631X (2023) https://doi.org/10.1117/12.3007662
Event: Applied Optics and Photonics China 2023 (AOPC2023), 2023, Beijing, China

Abstract

Currently, vocal cord lesion diagnosis of laryngoscopic images mainly relies on physicians' expertise and clinical experience. This greatly increases the work pressure of physicians and has limited efficiency. To solve the above problems, this study aims to construct a deep network structure named VCLR-Net based on the improved YOLOv7 to achieve the detection and recognition of vocal cord lesions. First, Convolutional Block Attention Modules (CBAM) are added to the HEAD network to improve the focus of color and spatial features on lesions. Next, the Alpha Intersection over Union loss (AlphaIOU) loss function is used to improve the robustness of the lesion recognition model. In the experimental results, the proposed VCLR-Net network achieves mAP and F1 of 0.762 and 0.748 in the image dataset. The network enables accurate lesion recognition for a large number of laryngoscopic images.

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

Citation Download Citation

Yanda Wu, Yuqing He, Dongyan Huang, Yang Liu, Jingxuan Zhu, and Hengli Zhang "Efficient vocal cord lesion recognition by combing Yolov7 and attention module", Proc. SPIE 12963, AOPC 2023: Optical Sensing, Imaging, and Display Technology and Applications; and Biomedical Optics, 129631X (18 December 2023); https://doi.org/10.1117/12.3007662

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