Pedestrian detection method and device

Abstract

The embodiment of the present invention provides a pedestrian detection method. The method comprises the steps of receiving a to-be-detected picture; adopting a neural network to determine the pedestrian information in the to-be-detected picture, wherein the pedestrian information comprises the pedestrian positions and the key point positions of the pedestrians; carrying out the postprocessing on the pedestrian information in the to-be-detected picture to obtain a pedestrian detection result. Therefore, according to the embodiment of the present invention, for the to-be-detected picture, the neural network can be adopted to obtain the pedestrian positions and the key point positions of the pedestrians, so that by introducing the key point information of the pedestrians, the shielding and overlap problem which is difficult to solve in the conventional pedestrian detection method is solved, and the detection accuracy is improved.

Description

technical field [0001] The present invention relates to the field of image recognition, and more particularly to a pedestrian detection method and device. Background technique [0002] Pedestrian detection is a subproblem of general object detection that has attracted a lot of interest from both academia and industry. First of all, pedestrian detection is a basic and important technology that can be applied in many fields such as security, automatic (assisted) driving, intelligent monitoring and human-computer interaction. [0003] Existing pedestrian detection cannot solve problems such as occlusion and overlapping of human bodies. Occlusion means that the target is only partially visible due to obstacles. The common occlusion result is that only the upper (lower) or left (right) half of the human body is visible. Overlapping human bodies means that the positional relationship of two or more people in the picture overlaps. The common overlapping of human bodies is more co...

AI Technical Summary

Problems solved by technology

In the case of overlap, non-maximum suppression can also cause detections for different objects to be mis-merged

Images