Research on Object Detection Method of Underground Video Image Based on SSD

Monitoring video image target detection in coal mine is of great significance to the safety of underground workers. In order to solve the problem of huge task and low efficiency of manual monitoring target, this paper establishes a deep learning model for target detection of underground images. Firstly, the deep neural network is trained by a large number of underground monitoring images, and then different deep learning algorithms are used to detect the target in the image. Finally, the mAP, precision and recall of different neural network target detection are calculated and evaluated, and detection effects of different deep learning detection algorithms are compared by analyzing the detection results. The analysis results show that the four deep learning models in this study have achieved good average accuracy. The target detection effect based on these four deep learning models is more accurate and efficient than other traditional target detection algorithms, which can be applied to target detection in the coal mines.