Crowd density analysis and detection positioning method based on density map

Abstract

The invention discloses a crowd density analysis and detection positioning method based on a density map. The method comprises the steps of preprocessing crowd images in a data set for training; converting the crowd images into a two-dimensional crowd density image by using a Gaussian filter; secondly, designing a deep separation hole convolution network model to learn a feature mapping function between an input image and a crowd density image; achieving an end-to-end prediction model, enabling the pixel values of the prediction density image to be subjected to integral summation to realize crowd counting, and finally, inputting the predicted crowd density image into a RetinaNet target detection network to achieve human head detection and positioning. According to the invention, classification of human heads and non-human heads in a crowd highly dense scene is realized, and the problems that specific positioning cannot be provided for crowd density analysis in the highly dense scene and missed detection exists in crowd detection can be solved.

Description

technical field [0001] The invention relates to a crowd detection and positioning method based on density map classification, which mainly designs a deep separation hole convolutional neural network model to generate a high-quality crowd density map and inputs the crowd density map into the target detection network to better realize target classification , to achieve crowd density analysis and head detection and positioning, which belongs to the cross-application field of image processing, target detection and artificial intelligence. Background technique [0002] The purpose of crowd density analysis and head detection and positioning is to obtain the spatial distribution information of people and detect the specific positioning information of people through the crowd density map. It has become a hot issue in the field of computer vision. Wide range of applications. There are three main methods for crowd density analysis: detection-based methods, regression-based methods, ...

AI Technical Summary

Problems solved by technology

This method combines the crowd density image with the target detection network, which solves the problem that the simple crowd counting cannot provide the spatial information of the crowd distribution, the specific location information of the characters, and the leakage caused by the small-sized head when the target detection network directly detects the crowd image. check question

Images