People flow counting method and system

Abstract

The invention provides a people flow counting method and system, and belongs to the technical field of image processing. The method comprises the following steps of: detecting the object of the imageset with the frame sequence according to the frame sequence; when a first image with a first object set is detected in the image set, obtaining a first local vector corresponding to the position of the first object set relative to the first image in the first image, and then obtaining a second local vector corresponding to the position of the second object set relative to the second image in the second image with the second object set in the image set; and obtaining a tracking value set according to the first local vector and the second local vector in combination with a preset tracking mapping relationship, and obtaining a tracking state set of objects in the first object set according to a relationship between each tracking value in the tracking value set and a preset tracking thresholdcondition. According to the invention, detection and tracking of high on-call rate are realized, and a high-precision and high-robustness people flow counting system is realized based on the detectionand tracking of the high on-call rate.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to an object tracking method, a method for counting objects using the object tracking method, a system for counting objects using the object tracking method, a device for counting objects and A computer-readable storage medium. Background technique [0002] With the continuous maturity and implementation of computer vision counting, the understanding of human behavior in real scenes has increasingly become the focus of users. Taking security scenes as an example, counting the characteristics of people passing by has become its core function; in addition, retail, supermarkets and stations, etc. There is also a demand for the function of counting passenger flow in the scene, so as to feedback the regularity of customers arriving at the store, and to count the queuing time of the checkpoint. In the actual people counting project, there are often requirements that the solution...

AI Technical Summary

Problems solved by technology

[0004] After the implementation of the deep learning people flow counting scheme, the effect of picture detection and tracking in actual scenes has been greatly improved, and a detection and tracking scheme using convolutional neural networks has emerged. To track the detection frame, although the counting accuracy has been improved, the calculation amount is much larger and the cost is significantly increased

[0005] Generally, people counting schemes based on deep learning have thorny technical problems such as error detection and repeated or invalid counting; error detection problems are usually caused by changes in human body posture, environmental occlusion, or scene asymmetry, and may be detected in real-time video images. The phenomenon of "flickering" is formed. The flickering phenomenon means that a detected object is detected as another object after passing through a posture change, an occluded environment or an asymmetric scene; the problem of counting repetition or invalidity is usually due to changes in human body posture and environmental occlusion during the tracking process. or caused by scene asymmetry, and may form a temporary "missing" phenomenon in the real-time detection video screen. The missing phenomenon means that an object being tracked is considered to be tracked but completed after passing through a posture change, an occluded environment or an asymmetric scene. Objects that actually continue to appear in subsequent frames in the frame and may be considered another tracked object

Images