Related Experiment Video
Updated: May 6, 2026
08:46
Using Synchrotron Radiation Microtomography to Investigate Multi-scale Three-dimensional Microelectronic Packages
Published on: April 13, 2016
10.0K
Lightweight Substation Equipment Defect Detection Algorithm for Small Targets
Jianqiang Wang1, Yiwei Sun2, Ying Lin2
1Department of Electronic and Communication Engineering, North China Electric Power University, Baoding 071003, China.
Sensors (Basel, Switzerland)
|September 28, 2024
Summary
A new Efficient Attentional Lightweight-YOLO (EAL-YOLO) algorithm improves substation equipment defect detection accuracy and efficiency. This lightweight model excels at identifying small defects, making it ideal for resource-constrained devices.
Area of Science:
- Electrical Engineering
- Computer Vision
- Artificial Intelligence
Background:
- Substation equipment defect detection is crucial for operational maintenance but faces challenges with complex scenarios, small target detection, and algorithm complexity.
- Current mainstream algorithms struggle with high missed detection rates for small targets and reduced precision, hindering deployment on devices with limited resources.
Purpose of the Study:
- To propose an Efficient Attentional Lightweight-YOLO (EAL-YOLO) algorithm for detecting defects in substation equipment, focusing on small targets and lightweight design.
- To enhance detection accuracy and precision while reducing computational complexity for deployment on resource-constrained devices.
Main Methods:
- Optimized the model backbone using EfficientFormerV2 and integrated the Large Separable Kernel Attention (LSKA) mechanism into Spatial Pyramid Pooling Fast (SPPF) for improved feature extraction.
- Developed a novel Attentional scale Sequence Fusion P2-Neck (ASF2-Neck) to enhance the detection of small target defects.
- Introduced a Lightweight Shared Convolutional Head (LSCHead) module to facilitate deployment on resource-constrained devices.
Main Results:
- EAL-YOLO demonstrated a 2.93 percentage point accuracy improvement over YOLOv8n, achieving 92.26% mAP50 for 12 typical equipment defects.
- The algorithm significantly reduced Floating Point Operations (FLOPs) by 46.5% and parameters by 61.17% compared to YOLOv8s.
- Achieved superior detection accuracy compared to mainstream models while maintaining a lightweight architecture.
Conclusions:
- The proposed EAL-YOLO algorithm effectively addresses the challenges of small target detection and computational complexity in substation equipment defect detection.
- EAL-YOLO offers a promising solution for real-time, accurate, and efficient defect detection in substation environments, especially on devices with limited computational power.
