# Power Defect Detection with Improved YOLOv12 and ROI Pseudo Point Cloud Visual Analytics

## TL;DR

The paper introduces an improved YOLOv12 framework for detecting power equipment defects, using advanced modules and pseudo point cloud analytics for better accuracy and interpretability.

## Contribution

The novel framework combines PG-RA, LR-RELAN, and DSAF Loss with ROI pseudo point cloud analytics for enhanced defect detection and interpretability.

## Key findings

- The improved YOLOv12 achieves robust defect detection in complex backgrounds and varying illumination.
- ROI pseudo point cloud construction with SOR/ROR denoising improves interpretability of defect regions.
- The framework maintains real-time performance while improving detection accuracy and interpretability.

## Abstract

What are the main findings?
An improved YOLOv12 integrating PG-RA, LR-RELAN, and DSAF Loss achieves more robust power-equipment defect detection under complex backgrounds and varying illumination.ROI-based pseudo point cloud construction with SOR/ROR denoising provides clearer local structural patterns, improving the interpretability of defect regions beyond 2D bounding box outputs.

An improved YOLOv12 integrating PG-RA, LR-RELAN, and DSAF Loss achieves more robust power-equipment defect detection under complex backgrounds and varying illumination.

ROI-based pseudo point cloud construction with SOR/ROR denoising provides clearer local structural patterns, improving the interpretability of defect regions beyond 2D bounding box outputs.

What are the implications of the main findings?
The proposed framework offers a lightweight and deployable solution for intelligent power inspection, supporting faster and more reliable condition assessment in real-world maintenance workflows.The low-cost pseudo point cloud visual analytics pipeline suggests a practical path to enhance explainability in industrial defect detection.

The proposed framework offers a lightweight and deployable solution for intelligent power inspection, supporting faster and more reliable condition assessment in real-world maintenance workflows.

The low-cost pseudo point cloud visual analytics pipeline suggests a practical path to enhance explainability in industrial defect detection.

Power-equipment fault detection is challenging in real-world inspections due to subtle defect cues and cluttered backgrounds. This paper proposes an improved YOLOv12-based framework for multi-class power defect detection. We introduce a Prior-Guided Region Attention (PG-RA) module and design a Lightweight Residual Efficient Layer Aggregation Network (LR-RELAN). In addition, we develop a Dual-Spectrum Adaptive Fusion Loss (DSAF Loss) function to jointly improve classification confidence and bounding box regression consistency, enabling more robust learning under complex scenes. To support defect-oriented visual analytics and system interpretability, the framework further constructs Region of Interest (ROI) pseudo point clouds from detection outputs and compares two denoising strategies, Statistical Outlier Removal (SOR) and Radius Outlier Removal (ROR). A Python-based graphical prototype integrates image import, defect detection, ROI pseudo point cloud construction, denoising, 3D visualization, and result archiving into a unified workflow. Experimental results demonstrate that the proposed method improves detection accuracy and robustness while maintaining real-time performance, and the ROI pseudo point cloud module provides an intuitive auxiliary view for defect-structure inspection in practical applications.