Pedestrian detection method based on deep learning and multi-layer stimulation

Abstract

The invention discloses a pedestrian detection method based on deep learning and multi-layer stimulation. The method is used to mark a location, where a target appears in a video, after the surveillance video and the target which needs to be detected are given. The method specifically includes the following steps: S1, acquiring a pedestrian dataset used for training a target detection model, and defining an algorithm target; S2, modeling a location deviation and apparent semantics of the pedestrian target; S3, establishing a pedestrian multi-layer stimulation network model according to a modeling result in the step S2; and S4, using the detection model to detect the pedestrian location in a surveillance image. The method is suitable for use in pedestrian detection in the real video surveillance image, and has better effect and robustness when various complicated situations are faced.

Description

technical field [0001] The invention belongs to the field of computer vision, in particular to a pedestrian detection method based on deep learning multi-layer stimulation. Background technique [0002] Since the end of the 20th century, with the development of computer vision, intelligent video processing technology has received extensive attention and research. Pedestrian detection is one of the important and challenging tasks, and its goal is to accurately detect the location of pedestrians in video surveillance images. This problem has high application value in the fields of video surveillance and intelligent robots, and is the basis of a large number of advanced vision tasks. But again, this problem has great challenges. One is how to express the information of the target area; the other is how to unify the extraction of candidate areas and target classification into modeling and optimization. high demands. [0003] The general pedestrian detection algorithm is divid...

AI Technical Summary

Problems solved by technology

This method mainly has the following problems: 1) It is based on traditional visual features, which can only express lower-level visual information, but pedestrian detection tasks require the model to have high-level abstract semantic understanding capabilities; 2) Candidate regions The extraction and classification of features are not optimized for end-to-end learning; 3) The features extracted based on deep learning have not been combined with multiple layers of stimuli, and the target features are not abstract and rich enough

Images