Related Experiment Video
Updated: Jan 13, 2026
07:03
Medical-grade Sterilizable Target for Fluid-immersed Fetoscope Optical Distortion Calibration
Published on: February 23, 2017
8.0K
ACCYolo: Transmission equipment inspection image detection method based on multi-scale and occluded targets
Xi Chen1, Fulong Yao1, Rongbin Cui1
1School of Software, Shenyang University of Technology, Shenyang, China.
Plos One
|October 28, 2025
Summary
This study introduces ACCYolo, an enhanced AI model for inspecting power transmission equipment using drones. It improves detection of multi-scale and occluded targets, ensuring reliable power supply.
Area of Science:
- Electrical Engineering
- Computer Vision
- Artificial Intelligence
Background:
- Global electricity demand necessitates robust transmission infrastructure.
- Unmanned Aerial Vehicles (UAVs) enhance inspection efficiency and safety.
- Harsh environments and complex terrains challenge UAV-based equipment inspection due to occlusion and scale variations.
Purpose of the Study:
- To develop an improved AI model for detecting multi-scale and occluded transmission equipment in UAV imagery.
- To enhance the accuracy and reliability of power infrastructure monitoring.
Main Methods:
- Proposed ACCYolo model based on YOLOv10n architecture.
- Integrated Acmix model with self-attention for occlusion handling.
- Utilized GELAN structure with Programmable Gradient Information (PGI) and ASFF module for multi-scale detection.
Main Results:
- ACCYolo demonstrated significant advantages in transmission equipment monitoring.
- Achieved an overall mean Average Precision (mAP@50) of 0.950.
- Effectively addressed challenges of occlusion and multi-scale targets.
Conclusions:
- ACCYolo provides an effective solution for UAV-based transmission equipment inspection.
- The model enhances operational efficiency and safety in power infrastructure maintenance.
- Contributes to ensuring the reliability of the power supply through advanced monitoring.
