Crowd counting method based on semi-supervised manifold embedding

Abstract

The invention discloses a crowd counting method based on semi-supervised manifold embedding. The crowd counting method comprises a training stage and a testing stage. In the training phase, firstly, training samples are randomly selected from a crowd data set; features are extracted, the training samples with the marks and without the marks are transformed from a feature space to a label space through a semi-supervised manifold embedding model, and a linear transformation function in label transformation is learned by utilizing structural information between adjacent images of the feature space and the semi-supervised manifold embedding model to obtain a trained semi-supervised manifold embedding model; and in the testing phase, in the trained semi-supervised manifold embedding model, a test sample is mapped from a feature space to a label space by using a linear transformation function learned in a training stage to obtain a matrix in the label space, the matrix represents probabilitydistribution of the sample in a corresponding category, and the maximum probability represents the population number of the sample.

Description

Technical field [0001] The invention belongs to the technical field of image processing and analysis methods, and relates to a crowd counting method based on semi-supervised manifold embedding. Background technique [0002] Crowd counting is a very attractive computer vision technology, which can count the number of people in an image through related image processing technology. It has broad application prospects in the fields of security, public resource management, and transportation assistance. Current methods usually only pursue accuracy, thus ignoring the time requirements of real-time applications. In addition, most of the existing technologies use supervised learning methods, which require a large amount of accurate labeling of data. However, in the actual application of a large amount of data, only a small part of the data is labeled. To label these large amounts of data requires a very large human resource overhead. At the same time, due to the complexity and length of...

AI Technical Summary

Problems solved by technology

[0007] The purpose of the present invention is to provide a crowd counting method based on semi-supervised manifold embedding, which solves the problems of slow crowd counting methods and the need for a large amount of labeled data in the prior art

Images