Crowd-based image recognition method, system, device, and storage medium

Abstract

The invention provides a crowd-based image recognition method, a system, a device and a storage medium. The method comprises the following steps: carrying out the joint calibration of a laser radar and at least one camera driven by a table device based on a monitoring region; when the camera rotates to each optical axis position, obtaining an image obtained by the camera corresponds to a local area of the monitoring area; a laser radar collects point cloud data of a monitoring area in real time, and identifies the real-time position of a point cloud cluster representing each pedestrian in the point cloud data; performing pedestrian image partition recognition on a current image shot by a camera based on an optical axis position through a first neural network, and performing matching according to the position of the pedestrian in the current image and a point cloud cluster of the pedestrian in a local area corresponding to the current optical axis position; and only subjecting the image partitions which are not recognized by the second neural network to image recognition through the second neural network. The laser radar can be combined to avoid repeated scanning, and the monitoring effect is enhanced.

Description

technical field [0001] The invention belongs to the field of visual monitoring, and in particular relates to a method, system, device and storage medium for image recognition based on crowds. Background technique [0002] The monitoring system is also called closed-circuit television monitoring system (English: Closed-Circuit Television, abbreviated as CCTV). A typical monitoring system mainly consists of five parts: front-end audio and video acquisition equipment, audio and video transmission equipment, back-end storage, control and display equipment. Composition, in which the back-end equipment can be further divided into central control equipment and sub-control equipment. There are many ways to form the front-end and back-end equipment, and the connection between them (also called the transmission system) can be realized through coaxial cables, twisted-pair lines, optical fibers, microwaves, wireless and other ways. [0003] In areas that need to be monitored, such as s...

AI Technical Summary

Problems solved by technology

However, due to the different speeds of pedestrians walking, pedestrian occlusions often occur in the shooting settings, resulting in missed detection or repeated scanning. On the one hand, setting up multiple devices increases the cost, and on the other hand, it is essential for pedestrians to occlude each other. There is not much improvement in the field, and many pedestrians in the images captured by multiple devices are repeated, which not only wastes the amount of calculation, but also makes it difficult to estimate the total number of pedestrians in the monitoring area, resulting in inaccurate data and weakening the monitoring. Effect

Images