Core Concepts
Enhanced lightweight YOLOv5 for transmission line detection.
Abstract
Abstract:
YOLOv5 challenges: computational load, detection accuracy.
Enhanced lightweight YOLOv5 for mobile devices.
Modules integration: C3Ghost, FasterNet.
wIoU v3 LOSS function for dataset imbalance.
Introduction:
Importance of transmission line monitoring.
Evolution from two-step to regression-based methods.
Methodology:
Replacement of modules in YOLOv5 with FasterNet and C3Ghost.
Introduction of WIoU loss function for dataset balance.
Experimental Results:
Dataset composition and division for training model.
Performance metrics: Precision, Recall, mAP@.5, mAP@.5-.95.
Conclusion:
Fostc3net model achieves improved accuracy and reduced parameters.
Ablation experiments show impact of structural changes on performance.
Stats
The proposed model achieves a 1% increase in detection accuracy, a 13% reduction in FLOPs, and a 26% decrease in model parameters compared to existing YOLOv5.
The number of parameters in the Fostc3net model is reduced by approximately 26.61%, and GFLOPs by about 13.09%.