A crowd density prediction method, device, storage medium and terminal based on multi-camera fusion

Abstract

The invention discloses a multi-camera fusion crowd density prediction method, device, storage medium and terminal. The method includes: multiple cameras collect scene images in real time and input them into a perspective target detection model and a pedestrian target detection model for detection, and generate detection objects Information and pedestrian target detection results; generate multiple camera perspective transformation matrices based on the detected object information; correct based on the pedestrian target detection results to generate pedestrian target body frame coordinates; call the perspective transformation matrix to calculate the perspective transformation results of the pedestrian target body frame coordinates , generate multiple perspective transformation results for fusion, and generate a plan view after deduplicating pedestrians in each overlapping area; according to the coordinates in the plan view, generate a crowd density heat map corresponding to the plan view. Therefore, by adopting the embodiment of the present application, a real-time planar crowd density thermal map is obtained through the fusion of perspective conversion results of multiple cameras at the same time, thereby improving the information perception capability and emergency response capability of the management department.

Description

technical field [0001] The present invention relates to the field of computers, in particular to a multi-camera fusion crowd density prediction method, device, storage medium and terminal. Background technique [0002] In recent years, unconventional emergencies have occurred frequently around the world, and these events are likely to bring catastrophic consequences, so the evacuation plan of large-scale groups has become an increasingly important issue in the field of public safety. So far, there have been a large number of model studies on the evacuation of large-scale crowds at the micro and macro levels, and these studies have strongly promoted the development of the field of emergency response based on unconventional emergencies. However, the uncertainties surrounding the evacuation of large populations during unconventional emergencies are not well understood. [0003] In the current crowd monitoring method, the real-time monitoring of ordinary cameras and the thermal...

AI Technical Summary

Problems solved by technology

However, the uncertainties surrounding the evacuation of large populations during unconventional emergencies are not well understood

[0003] In the current crowd monitoring method, the real-time monitoring of ordinary cameras and the thermal map imaging monitoring of infrared cameras are usually used to monitor different situations in application scenarios, because these two methods can only monitor the distribution of crowds in a single range , cannot macroscopically display the accurate distribution of crowd density in the venue, so that the management department cannot macroscopically observe the crowd distribution in the entire scene, thereby reducing the information perception ability and emergency response ability of the management department

Images