Real-time insulator defect detection is critical for ensuring the reliability and safety of power transmission systems. However, deploying deep learning models on edge devices presents significant challenges due to limited computational resources and strict latency constraints. To address these issues, we propose YOLOLS, a lightweight and efficient detection model derived from YOLOv8n and optimized for real-time edge deployment. Specifically, YOLOLS integrates GhostConv to generate feature maps through stepwise convolution, reducing computational redundancy while preserving representational capacity. Moreover, the C2f module is restructured into a ResNet–RepConv architecture, in which convolution and Batch Normalization layers are fused during inference to reduce model complexity and enhance inference speed. To further optimize performance, a lightweight shared-convolution detection head significantly reduces parameter count and computational cost without compromising detection accuracy. Additionally, an auxiliary bounding box mechanism is incorporated into the CIoU loss function, improving both convergence speed and localization precision. Experimental validation on the CPLID dataset demonstrates that YOLOLS achieves a 42.4% reduction in parameters and a 48.1% decrease in FLOPs compared to YOLOv8n while maintaining a high mAP of 91%. Furthermore, when deployed on Jetson Orin NX, YOLOLS achieves 44.6 FPS, ensuring real-time processing capability. Compared to other lightweight YOLO variants, YOLOLS achieves a better balance between accuracy, computational efficiency, and inference speed, making it an optimal solution for real-time insulator defect detection in resource-constrained edge computing environments.

Transmission-line inspection requires a single detector to localize complete insulator strings together with much smaller broken-shell and flashover-damaged regions. These targets differ in spatial extent and in their dependence on local detail and surrounding context, which complicates cross-scale feature fusion when accurate bounding boxes are required. This study develops receptive-field-aware path aggregation (PRA) and bidirectional receptive-field-aware aggregation (BRA), combining path aggregation with receptive-field expansion and adaptive scale weighting. Compared with the path aggregation network (PANet), PRA increases mean average precision (mAP) at an intersection-over-union (IoU) threshold of 0.5 (mAP@0.5) by 2.2 percentage points and mAP averaged over IoU thresholds from 0.5 to 0.95 (mAP@0.5: 0.95) by 4.6 points; BRA produces gains of 1.5 and 4.0 points, respectively. The highest mAP@0.5:0.95 of 91.3% is jointly achieved by Swin-T-PRA + Alpha-CIoU and Swin-T-BRA + Alpha-CIoU, with corresponding mAP@0.5 values of 98.6% and 98.5% and model-only inference speeds of 48.5 and 48.3 FPS on an RTX 3080 Ti, respectively. The larger gains across stricter IoU thresholds indicate that PRA/BRA provide greater benefits when more stringent box-overlap criteria are imposed, although the present evaluation does not independently isolate the bounding-box regression mechanism.