Multi-target detection and tracking method and system, storage medium and application

Abstract

The invention belongs to the technical field of target detection and tracking, and discloses a multi-target detection and tracking method and system, a storage medium and application; a laser radar scans to obtain point cloud data, transmits the collected data to a server side, and uploads the data to a cloud storage; after radar data are obtained, step points are extracted through moving point extraction, point clustering and random forest; track matching is used for footstep tracking; after the data is processed and the footstep information of the pedestrian is obtained, a processing result is sent to the user in a visual form; meanwhile, when the pedestrian is detected, relevant walking information is output and stored in a file, and the information comprises walking speed and footstep length information. Compared with pedestrian tracking based on a visual sensor, tracking based on the laser radar is not interfered by the environment, the algorithm is easy, and the method is suitable for being used in a multi-person scene. Verification is carried out in a laboratory environment and a real home environment of the elderly, and an ideal effect can be achieved.

Description

technical field [0001] The invention belongs to the technical field of target detection and tracking, and in particular relates to a multi-target detection and tracking method, system, storage medium and application. Background technique [0002] Present: In the field of computer research, object detection and tracking are important and fundamental tasks. At present, there are many applications of computer-related technologies in detecting targets, and they have applications in various industries, such as image recognition, vehicle detection, and intelligent monitoring. There are also many solutions for target tracking, among which computer vision detection is more commonly used. The detection scheme based on visual sensors is more intuitive and accurate, and its development is relatively mature. But it also has many shortcomings, such as being greatly affected by light, it is difficult to distinguish multi-target objects mixed together, and the algorithm consumes more res...

AI Technical Summary

Problems solved by technology

[0013] (1) At present, the pedestrian tracking technology in the computer field has problems such as how to deal with complex monitoring environments, how to adopt effective detection algorithms for high-frequency continuous detection, and improve the reliability and accuracy of detection and tracking.

[0014] (2) At present, the pedestrian tracking technology in the computer field has monitoring and recognition based on computer vision, which will cause a series of social problems such as privacy leakage; and the cost of real-time monitoring is usually very high, for example, using a set of lens sensors requires a lot of overhead

The second is that the detection target is more seriously blocked.

In many scenes, some footstep points are very short or sparse, which greatly increases the error of random forest in judging human feet. These clusters have basically lost some characteristics that a person's footsteps should have, such as footsteps the length of

Even worse, because the number of scanned footsteps is small or too sparse, clustering will ignore them

Images