Abstract: In order to solve the low detection precision issue of unsafe personnel state around coal mine belt conveyor due to variable scale and background occlusion, a novel DSP-YOLO intelligent fast detection algorithm is proposed. Firstly, to address the problem of insufficient diversity of original data and limited model generalization ability, the traditional data augmentation is improved with the GridMask data augmentation strategy to increase the robustness of the detection algorithm for the occluded or small targets in a complex background. Secondly, to address the requirements of the detection precision and efficiency of unsafe personnel state, the Shift-wise convolution is introduced into the backbone network to extend the receptive field and enhance the global feature extraction ability. Meanwhile, the dynamic multi-scale feature fusion module is designed, making full use of the detail features and multi-scale features to improve the adaptability of the detection algorithm for the multi-scale targets and a complex background. And the PIoUv2 loss function is adopted to alleviate the problem of missing detection and false detection caused by the occluded targets and small targets. In addition, to further improve the detection performance of DSP-YOLO in an actual scene, the detection model is trained using the laboratory data and transferred it for the on-site data. The results show that the precision, mAP_0.5 and mAP_0.5:0.95 on the data set combined with the traditional data augmentation and the GridMask strategy increase by 3.4%, 1.9% and 6.6%, respectively, compared to the original data set; the precision, recall, mAP_0.5 and mAP_0.5:0.95 of the DSP-YOLO reach 94.8%, 89.4%, 93.2% and 70.5%, respectively, with an increase of 2.3%, 1.1%, 0.9% and 2.3% compared to the baseline model; for the on-site data, the detection precision reaches 99.1% by using the transfer application strategy. The research results show that the proposed new algorithm is capable of achieving the accurate and intelligent detection of unsafe personnel state around the coal mine belt conveyor.