Head detection based on YOLOv3 image processing technology

Yang Liu; Yunling Li; Keyi Yang; Xinyu Liu

doi:10.1117/12.2625717

9 December 2021 Head detection based on YOLOv3 image processing technology

Yang Liu, Yunling Li, Keyi Yang, Xinyu Liu

Proceedings Volume 12129, International Conference on Environmental Remote Sensing and Big Data (ERSBD 2021); 1212909 (2021) https://doi.org/10.1117/12.2625717
Event: 2021 International Conference on Environmental Remote Sensing and Big Data, 2021, Wuhan, China

Abstract

With the advancement of science and technology, image processing technology is widely used. In this paper, we use computer technology to process the image data of the screenshots in the classroom monitoring to achieve the goal of extracting human heads. In this study, we selected screenshots of different size classrooms, camera views, and classroom lighting conditions. Combined with the PASCAL VOC dataset format, we applied the Labellmg-master image annotation tool to annotate the human head. The traditional YOLOv3 algorithm misses and repeats detections. Its accuracy and recall rates are poor. Therefore, we propose an optimization algorithm for classroom monitoring via head detection based on the YOLOv3 algorithm. The main improvements are the clustering the dataset labels with the Kmeans algorithm, increasing the size of the output feature map, and optimizing the loss function. Experiments were conducted with both the YOLOv3 and optimized YOLOv3 algorithms. By comparing the experimental results, we can conclude that the optimized YOLOv3 algorithm can improve the detection accuracy by 6% and recall rate by 11% on the test set, which has a high detection performance. This shows that image data processing technology has a better application effect in classroom monitoring.

Citation Download Citation

Yang Liu, Yunling Li, Keyi Yang, and Xinyu Liu "Head detection based on YOLOv3 image processing technology", Proc. SPIE 12129, International Conference on Environmental Remote Sensing and Big Data (ERSBD 2021), 1212909 (9 December 2021); https://doi.org/10.1117/12.2625717

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