Deep learning-based personnel retrograde running detection method in electronic escalator running process

The invention discloses a deep learning-based personnel retrograde driving detection method in an electronic escalator operation process, which comprises the following steps of: inputting a collected personnel image on an electronic escalator into a target identification network model to obtain continuous multi-frame personnel position information; damping, filtering and matching are carried out on continuous multi-frame personnel position information, and personnel tracking information of continuous multi-frame images is obtained; and in combination with the personnel tracking information of the continuous multi-frame image and the identified running direction of the electronic escalator, a retrograde moving sample range is obtained, and the movement direction of the personnel is judged based on the retrograde moving sample range. According to the method, the deep learning technology is utilized, the image data in the operation process of the electronic escalator is learned, an algorithm and a system framewo