An Improved YOLOv8 Based Unsafe Behavior Detection Algorithm for Coal Mine Underground Personnel

Aiming at the problems of high computational cost and poor detection effect in behavioral target recognition in coal mine underground, in this paper, a high performance target detection method based on YOLOv8n is designed. First, the convolutional block RFCAConv with the perceptual field attention mechanism is used instead of the convolutional block, By efficiently capturing spatial information within the sensory field, the overall performance of the model is significantly improved; the SEAttention module is introduced, which utilizes the spatial attention mechanism to make the model focus more on the identification of important targets; the experimental results show that, compared with the commonly used models, such as YOLOv5s, Faster R-CNN, SSD, etc., the performance of the improved model has a significant improvement. The experimental results show that compared with the commonly used models such as YOLOv5s, Faster R-CNN, SSD, etc., the performance of the improved model is significantly improved. The <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{mAP} {@} 0.5$</tex> and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{mAP} {@} 0.5: 0.95$</tex> were increased by 2.7 % and 3 %, with an FPS value of 5.8 ms. The algorithm effectively reduces the number of parameters while improving the performance, and meets the requirements of accuracy and timeliness.