Group migration counting method and system

Abstract

The embodiment of the invention provides a group migration counting method and system, a multi-order feature extraction module is adopted to enhance the resolution of feature extraction in different visual counting scenes, and a segmentation fusion module is adopted to perform classification counting and regression counting according to local features. Whether the local features of each calculation stage should be segmented is determined. A background domain migration model adapted to group counting is created by classification and regression counting; the basic thought that segmentation is performed along multi-order features and counting is performed aiming at local features is adopted, so that the generalization ability of the model aiming at multiple application scenes is realized. The anti-interference performance of the model under extreme conditions is improved, and the defect that a current mainstream visual counting model only uses a single convolutional network to realize visual counting is overcome.

Description

Technical field [0001] This application embodiment relates to the field of computer technology, and more particularly to a group migration count method and system. Background technique [0002] With the rise of video surveillance, medical devices, intelligent manufacturing and smart cities, more and more researchers focus on their research perspective in the study of group counts. Both the camera use the camera as the sensor to achieve the group count by obtaining the visual information of the group scene. A large number of pre-experiments have been confirmed that the group count model plays an important role in social security and control management. [0003] At present, the research of group count model is mainly divided into four categories: based on detection counting model; regression count model; density estimation count model and density estimation count model based on convolutional neural network. The test counting model is mainly applied to the visual count. In the scene...

AI Technical Summary

Problems solved by technology

[0006] The main purpose of the embodiment of the present application is to propose a method and system for group migration counting, which solves the problem that the learning effect of the group counting model based on the convolutional neural network is different when the learning background environment in the prior art is different from the background environment of the closed set. The problem that the effect of learning in multiple contexts will decrease

Images