High-precision low-delay crowd distribution calculation method for large indoor venue

Abstract

The invention provides a high-precision low-delay crowd distribution calculation method for a large indoor venue. The method comprises the following steps: step 1, carrying out scene gridding; step 2, carrying out equipment deployment; step 3, collecting and processing sample data; step 4, designing a recurrent neural network (RNN); step 5, setting a calculation and training period; step 6, training a recurrent neural network (RNN); and step 7, calculating crowd distribution based on a recurrent neural network (RNN). The crowd distribution obtained by the method is very close to the real distribution, and the method has practical value.

Description

technical field [0001] The invention relates to a high-precision and low-delay large-scale indoor venue crowd distribution calculation method. Background technique [0002] With the rapid development of the economy, large and complex modern buildings emerge one after another, such as exhibition venues, sports venues, transportation hubs and commercial complexes, which have become the business card of urban development. These venues often have large passenger flow and dense crowds. Due to the large area and complex structure of the venue, it is difficult to perceive the distribution of the crowd quickly and accurately, resulting in untimely or inaccurate operation command, which leads to adverse events such as crowd flow blockage or stampede. The potential safety hazards have brought great pressure to the operation and management of venues. In order to solve the above problems, various crowd positioning technologies have been introduced by the technical community. The curren...

AI Technical Summary

Problems solved by technology

[0003] In the current engineering practice, the above two solutions based on image recognition and radio frequency positioning cannot well meet the requirements of crowd positioning in large indoor venues

Among them, the positioning scheme based on image recognition mainly calculates the number and position of people in the picture by performing image recognition on the images in video surveillance, and then obtains the distribution of people in the monitoring area. This scheme is characterized by high precision but poor real-time performance. , especially in such large venues, a single positioning calculation takes a long time, ranging from half an hour to several hours, which is far from meeting user requirements; while radio frequency-based positioning solutions usually use Bluetooth, WiFi or zigbee And other low-power wireless positioning technology, deploy a certain density of wireless communication base stations in the venue, and at the same time, the people who pre-locate need to carry matching positioning tags or bracelets and other equipment. The wireless base station equipment locates by collecting the wireless signals sent by each tag device. The real-time performance is good, but due to the large area of ​​the venue and the large flow of people, the implementation of the scheme requires a large number of base stations and tags, and the one-time investment and operation and maintenance costs are high, which is not conducive to popularization and promotion, so it is not suitable

Images