Research on Mine Target Detection Method Based on Improved YOLOv8

Aiming at the complex environment such as uneven illumination and variable target scales in mines, achieving precise and realtime detection of potential collision targets such as personnel and equipment by autonomous mining vehicles is crucial for improving mine safety and work efficiency. To address the above problems, this paper proposes an improved algorithm based on YOLOv8. The Robust Feature Downsampling (RFD) module is introduced to solve the problem of small target feature loss, and the weighted Bidirectional Feature Pyramid Network (BiFPN) is adopted to enhance the multi-scale feature fusion capability. Experimental results show that the improved model achieves detection accuracies of 91.5% mAP and 86.1% F1-score on the self-built mine scene dataset, which are 3.3% and 2.6% higher than the original benchmark model YOLOv8-n, respectively. Meanwhile, the model complexity is significantly reduced. This algorithm ensures the efficient and accurate operation of the real-time obstacle detection system for unmanned loaders.