A Study of Energy Meter Miswiring Recognition Based on Image Enhancement and YOLO

For the issue of detecting wiring errors in electric energy meters, this paper proposes an automatic recognition method based on image enhancement and the YOLO model. With the development of smart grid, the safety risks caused by wiring errors in electric meters and the limitations of manual inspection have become increasingly prominent. To further improve recognition efficiency and accuracy, this study integrates deep learning techniques and optimizes the YOLOv5 object detection model to achieve a more efficient automated recognition of the wiring states of electric energy meters.A YOLOv5-based algorithmic framework for wiring error detection was constructed, which includes an Image Enhancement Module (Enhance Module) and an Intelligent Judgment Module (IJ Module). The Enhance Module employs the Retinex theory to decompose low-light images into illumination and reflectance components, enhancing image details and visibility through optimized reconstruction. The IJ Module sorts and matches the output results of the YOLO model, automatically determining the type of wiring errors.Experimental results demonstrate that, after image enhancement, the precision of the YOLOv5 model increased from 96 to 99%, the mean average precision (mAP@50) improved from 96 to 97%, and the mAP@50–95 showed a slight increase to 50%. Both subjective and objective evaluations confirm that image enhancement based on the Retinex theory effectively improves detection performance under low-light conditions. Combined with the YOLOv5 model, the method achieves efficient and accurate identification of wiring errors in electric energy meters. This method effectively addresses issues such as insufficient lighting and complex wiring configurations in electric meter inspections.