YOLO-PR: Multi Pose Object Detection Method for Underground Coal Mine

Object detection in coal mines is a crucial task for developing artificial intelligence assisted supervision systems in mining operations. Due to the complex environment of the underground coal mine and the variability of object poses, the general object detection algorithms cannot provide good performance. Hence, an improved underground multi-pose object detection method named YOLO-PR has been proposed. An EPA mechanism is designed to improve the feature extraction and representation capabilities of the backbone network, thereby enhancing the model's adaptability to pose variations. The RFB modules are integrated into the neck network to enhance its ability to learn multi-scale features. And the original loss function is optimized to better suit low-quality images in complex underground scenarios, thereby enhancing detection accuracy and stability. Experimental results demonstrate that, compared to the baseline model, the proposed YOLO-PR method achieves improved average detection accuracy for multi-pose objects while introducing only a marginal increase in model parameters. And, detection speed is maintained. When compared to other general detection models, YOLO-PR exhibits superior performance in completing multi-pose object detection tasks in the complex underground environments of coal mines.