Advances in Computer Vision and Sensor-Based Methods for Intelligent Power Transmission Line Inspection

Area of Science:

• Electrical Engineering
• Computer Science
• Artificial Intelligence

Background:

• Electricity is crucial for modern infrastructure, but transmission line inspection is costly and labor-intensive.
• Unmanned Aerial Vehicles (UAVs), advanced sensors, and deep learning-based Computer Vision are transforming inspection methods.

Purpose of the Study:

• To systematically review Computer Vision applications in electric power transmission line inspection.
• To synthesize findings across monitoring platforms, vision tasks, datasets, and industrial adoption.

Main Methods:

• Systematic literature review of 148 studies (2018-2026) from 1493 Scopus records.
• Screening and quality scoring based on PRISMA guidelines.
• Synthesis across four key axes: platforms/sensors, CV tasks, datasets/metrics, and adoption.

Main Results:

• You Only Look Once (YOLO) models dominate real-time edge deployment.
• Transformer architectures are increasingly adopted for inspection tasks.
• Persistent gaps exist in dataset availability, domain generalization, and field validation.

Conclusions:

• Computer Vision, particularly with UAVs, offers significant potential for efficient power line inspection.
• Addressing challenges in data and validation is crucial for achieving fully autonomous inspection systems.
• Future research should focus on robust, sustainable, and field-validated autonomous inspection solutions.