Insulator defect detection in power inspection image using focal loss based on YOLO v4

Xin Hu; Yunqiang Zhou

doi:10.1117/12.2626694

16 December 2021 Insulator defect detection in power inspection image using focal loss based on YOLO v4

Xin Hu, Yunqiang Zhou

Proceedings Volume 12153, International Conference on Artificial Intelligence, Virtual Reality, and Visualization (AIVRV 2021); 121530E (2021) https://doi.org/10.1117/12.2626694
Event: International Conference on Artificial Intelligence, Virtual Reality, and Visualization (AIVRV 2021), 2021, Sanya, China

Abstract

With the continuous expansion of the power industry, insulators, as one of the important power transmission equipment, have been exposed for a long time, causing wear, string loss and other problems, which threaten the safe operation of the power system. In this paper,yolov4 target detection algorithm based on deep learning is used. Based on CSPDarknet-53 and adding SPP Net, the receptive field is greatly increased. PANet is introduced to strengthen the feature extraction network，and focal loss is used as a function to calculate the loss.Then the models trained using RetinaNet and yolov4 are compared.The experimental data show that the defect detection accuracy can reach 96.2% and the recall rate can reach 97.7%.

Citation Download Citation

Xin Hu and Yunqiang Zhou "Insulator defect detection in power inspection image using focal loss based on YOLO v4", Proc. SPIE 12153, International Conference on Artificial Intelligence, Virtual Reality, and Visualization (AIVRV 2021), 121530E (16 December 2021); https://doi.org/10.1117/12.2626694

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