Convolutional neural network based crowd density distribution estimation method

Abstract

The invention relates to a convolutional neural network based crowd density distribution heat map generation method, which comprises the steps of dividing a crowd picture set into a training sample set and a test sample set, performing crowd label image segmentation by using convolutional neural network architecture, and carrying out number regression by using a convolutional neural network; correcting a density map through a multi-scale template operation, generating a crowd density distribution heat map according to the corrected density map and the regression number, and completing crowd density distribution estimation. According to the invention, deep characteristics of an image are extracted by using a powerful learning ability of the full convolutional neural network so as to perform accurate crowd segmentation, and low efficiency and blindness of density calculation of a traditional method for full image characteristics are overcome; a crowd near-far perspective effect is overcome to a certain extent through multi-scale template correction; and mapping is performed in allusion to the estimated number of people, lateral comparison can be performed on heat maps of different cameras, the method is applicable to various crowd scenes, and the crowd density distribution heat map can be acquired in real time.

Description

technical field [0001] The invention relates to the technical field of crowd density distribution estimation, in particular to a method for crowd density distribution estimation based on a convolutional neural network. Background technique [0002] Crowd density in public places is a key indicator related to passenger flow and safety issues in public places. With the development of intelligent monitoring technology, the research on the method of directly generating crowd density heat map from camera video data has important practical significance and social value. Traditional crowd density distribution methods mainly include two categories: 1. methods based on foreground segmentation, feature extraction and population regression based on motion information in video surveillance; A method for estimating people from different but complementary sources and finding the total number of people in a picture. The above traditional methods can achieve better detection results under ...

AI Technical Summary

Problems solved by technology

In the prior art, the crowd density estimation method based on texture features such as gray-level co-occurrence matrix is ​​to calculate the gray-level co-occurrence matrix of training samples in multiple directions of the image, and calculate its corresponding metrics such as contrast, uniformity, entropy, etc. Characterized as a feature vector, and then use methods such as principal component analysis, and through linear regression analysis, to establish a linear regression equation to estimate the sample to be tested. It is easy to misjudge uninhabited areas. At the same time, when the crowd density is low, it cannot Get effective texture information, unable to make correct estimation of low-density areas

Images