Experimental monkey indoor fighting behavior detection method based on target tracking

Through the combination of multi-objective tracking and interpolation algorithms, considering individual velocity and movement direction, identifying fighting behaviors in the experimental monkey room, solving the problem of high misjudgment rate in the existing technology, and achieving more accurate and widely applicable fighting behavior detection.

CN118942111APending Publication Date: 2024-11-12GUANGZHOU CITY CONSTR COLLEGE
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202410965399.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-18
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

The prior art has a high misjudgment rate when identifying fighting behaviors, especially due to the lack of speed factors and insufficient video clarity.

Method used

Multi-objective tracking algorithm and interpolation algorithm are used to judge fighting behaviors through changes in individual velocity, acceleration, motion direction and spacing, including the recognition conditions of chasing and entanglement fighting behaviors, and use spline interpolation or cubic polynomial interpolation to calculate the individual's continuous trajectory. .

Benefits of technology

It improves the accuracy of combat behavior recognition, reduces the misjudgment rate, and expands the scope of application scenarios, so as to effectively identify combat behaviors under lower video clarity.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118942111A_ABST
    Figure CN118942111A_ABST
Patent Text Reader

Abstract

The invention discloses an experimental monkey indoor fighting behavior detection method based on target tracking, and the method comprises the following steps: 1), carrying out the individual recognition and tracking of an experimental monkey through a multi-target tracking algorithm, and obtaining the mass center of each individual in each frame; 2) obtaining a continuous trajectory of the individual by using an interpolation algorithm, and further obtaining a position (x, y, t), a speed (v, t), an acceleration (a, t) and a movement direction (alpha, t) of the mass center of the individual along with time change; and 3) according to the calculated speeds (v, t) of the individuals, the accelerated speeds (a, t) of the individuals, the distances (d, t) between the individuals and the included angles (delta alpha, t) of the movement directions between the individuals, judging whether a fighting behavior exists or not. According to the method, the recognition effect on the fighting behavior can be improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a fighting behavior detection method. Background Art

[0002] Currently, video-based surveillance technology is booming, and fighting behavior is a hot topic in the algorithm for realizing target behavior recognition. In farms where experimental animals such as experimental monkeys are kept, how to accurately and quickly identify their fighting behavior is of great significance to the experimental animal breeding industry, the vaccine industry, and public health.

[0003] In the prior art, a Chinese patent application with application number 201610563808.X and invention name "A method and device for intelligent golden eye identification of people gathering and fighting alarm" (hereinafter referred to as prior art one) discloses a method of judging whether there is gathering behavior, falling behavior, and fighting behavior based on the distance between people. This method does not consider the speed factor. People walking slowly in the video may be judged as gathering behavior, so the misjudgment rate is high. Another Chinese patent application (hereinafter referred to as prior art two), application number 202211035042.X, invention name "A method for identifying fighting behavior in a video surveillance system", discloses a method of identifying the human body frame and judging whether a fighting behavior occurs by the joint angle, but when the human body is far away from the camera or the pixel is not high, it is difficult to identify the human body frame, and it is also difficult to identify the fighting behavior. Summary of the invention

[0004] The purpose of the present invention is to provide a detection method that can improve the recognition effect of fighting behavior.

[0005] The purpose of the present invention is achieved through the following technical solution: a method for detecting indoor fighting behavior of experimental monkeys based on target tracking, comprising the following steps:

[0006] 1) Use the multi-target tracking algorithm to identify and track individual experimental monkeys and obtain the center of mass of each individual in each frame;

[0007] 2) Use the interpolation algorithm to obtain the individual's continuous trajectory, and then further obtain the position (x, y, t), velocity (v, t), acceleration (a, t), and direction of movement (α, t) of the individual's center of mass over time;

[0008] 3) Determine whether there is fighting behavior based on the calculated individual speed (v, t), individual acceleration (a, t), distance between individuals (d, t), and angle between the movement directions of individuals (Δα, t).

[0009] Step 3) includes the identification of fighting behavior in which one person chases another person, and the following conditions must be met at the same time:

[0010] The velocities (v, t) of both bodies are greater than the threshold v0;

[0011] In the s seconds before and after the current time t, one or both body accelerations (a, t) are greater than the threshold a0;

[0012] The distance between individuals (d, t) is less than the threshold d0;

[0013] The angle (Δα,t) between the movement directions of the two bodies is less than the threshold α0.

[0014] Step 3) also includes the identification of mutually entangled fighting behaviors, which requires the following conditions to be met simultaneously:

[0015] The velocities (v, t) of both bodies are less than the threshold v1;

[0016] In the s seconds before and after the current time t, the accelerations of both bodies (a, t) are less than the threshold a1;

[0017] The distance between individuals (d, t) is less than the threshold d0;

[0018] The change per unit time of the movement direction (α, t) of the two bodies is greater than the threshold α1;

[0019] Among them, v0>v1, a0>a1, α0<α1.

[0020] It is recommended to calculate individual continuous trajectories by spline interpolation or cubic polynomial interpolation.

[0021] Beneficial effects:

[0022] Compared with the first prior art, the present invention comprehensively considers the individual's position, speed, acceleration, and movement direction information, which can more effectively prevent misjudgment and make the judgment more accurate; compared with the second prior art, the present invention only needs to achieve video clarity for tracking, without the need to identify individual frames, and has a wider range of application scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a logic flow chart of the indoor fighting behavior detection method of experimental monkeys based on target tracking of the present invention;

[0024] Figure 2 It is a logic flow chart of the present invention for identifying chasing and fighting behaviors and entanglement and fighting behaviors;

[0025] Figure 3 The track of the two individuals chasing and fighting identified by the method of the present invention;

[0026] Figure 4 A schematic diagram showing the distance between two points;

[0027] Figure 5 A schematic diagram showing the angle between the directions of motion of two points. DETAILED DESCRIPTION

[0028] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0029] The method for detecting indoor fighting behavior of experimental monkeys based on target tracking of the present invention is as follows Figure 1 As shown, the following steps are included:

[0030] 1) Use the multi-target tracking algorithm to identify and track individual monkeys and obtain the center of mass of each individual in each frame

[0031] The individual calibration frame, that is, the smallest box enclosing the individual, can be obtained through the multi-target tracking algorithm, and then the center of mass of the calibration frame can be found.

[0032] 2) Use the interpolation algorithm to obtain the individual's continuous trajectory, and then further obtain the position (x, y, t), velocity (v, t), acceleration (a, t), and direction of movement (α, t) of the individual's center of mass over time.

[0033] It is recommended to calculate individual continuous trajectories by spline interpolation or cubic polynomial interpolation.

[0034] 3) Determine whether there is fighting behavior based on the calculated individual speed (v, t), individual acceleration (a, t), distance between individuals (d, t), and angle between the movement directions of individuals (Δα, t).

[0035] Figure 2 It is a logic flow chart of the present invention for identifying chasing and fighting behaviors and entanglement and fighting behaviors.

[0036] In general, the identification of fighting behavior in which one person chases another person requires the following conditions to be met at the same time:

[0037] The displacement per unit time (speed) is large, and the speeds of both bodies (v, t) are greater than the threshold v0;

[0038] Sudden acceleration: in the s seconds before and after the current time t, one or both body accelerations (a, t) are greater than the threshold a0;

[0039] The distance between individuals is short, and the distance between individuals (d, t) is less than the threshold d0;

[0040] One body chases another body, and the angle between the movement directions of the two bodies (Δα,t) is less than the threshold α0.

[0041] To identify mutually entangled fighting behaviors, the following conditions must be met at the same time:

[0042] The displacement (speed) per unit time is not large, and the speeds of the two bodies (v, t) are both less than the threshold v1;

[0043] The acceleration is small. In the s seconds before and after the current time t, the accelerations of the two bodies (a, t) are both less than the threshold a1;

[0044] The distance between individuals is short, and the distance between individuals (d, t) is less than the threshold d0;

[0045] When fighting with each other, the movement directions of the individuals change suddenly, and the changes in the movement directions (α, t) of the two individuals per unit time are greater than the threshold α1.

[0046] A brief summary is shown in Table 1.

[0047] Table 1

[0048]

[0049] The thresholds in Table 1 are set according to the actual scene, and may vary due to different parameters such as device pixels, focal length, viewing angle, and venue size, and satisfy the following relationship: v0>v1, a0>a1, α0<α1.

[0050] Specifically, when judging, Figure 2 As shown, it is recommended to first determine whether the distance between individuals is less than the threshold d0, and then further determine the displacement per unit time, acceleration, and angle of motion direction.

[0051] A individual chases B individual in a fight, Figure 3 The trajectory of the two bodies chasing and fighting is obtained through tracking algorithms and interpolation, where points A and B are the initial moments, and points A' and B' are the current moments. The velocity (v, t) of the individual, the acceleration (a, t) of the individual, the distance between the individuals (d, t), and the angle between the movement directions of the individuals (Δα, t) are calculated through the trajectory, satisfying:

[0052] The velocities (v, t) of both bodies are greater than the threshold v0;

[0053] In the s seconds before and after the current time t, a body acceleration (a, t) is greater than the threshold a0;

[0054] The distance d' between individuals is less than the threshold d0, such as Figure 4 As shown;

[0055] The angle (Δα,t) between the moving directions of the two bodies is less than the threshold α0, such as Figure 5 shown.

[0056] It can be seen that the method of the present invention can identify the chasing and fighting between two individuals.

[0057] For two bodies entangled and fighting, the velocity (v, t) of the individual, the acceleration (a, t) of the individual, the distance (d, t) between the individuals, and the angle (Δα, t) between the movement directions of the individuals are calculated through the trajectory, which also meets the entanglement and fighting conditions of the present invention. It can be seen that the method of the present invention can also identify entanglement and fighting.

Claims

1. A method for detecting indoor fighting behavior of experimental monkeys based on target tracking, characterized in that: The steps include: 1) Use the multi-target tracking algorithm to identify and track individual experimental monkeys and obtain the center of mass of each individual in each frame; 2) Use the interpolation algorithm to obtain the individual's continuous trajectory, and then further obtain the position (x, y, t), velocity (v, t), acceleration (a, t), and direction of movement (α, t) of the individual's center of mass over time; 3) Determine whether there is fighting behavior based on the calculated individual speed (v, t), individual acceleration (a, t), distance between individuals (d, t), and angle between the movement directions of individuals (Δα, t).

2. The detection method according to claim 1, characterized in that: Step 3) includes the identification of fighting behavior in which one person chases another person, and the following conditions must be met at the same time: The velocities (v, t) of both bodies are greater than the threshold v0; In the s seconds before and after the current time t, one or both body accelerations (a, t) are greater than the threshold a0; The distance between individuals (d, t) is less than the threshold d0; The angle (Δα,t) between the movement directions of the two bodies is less than the threshold α0.

3. The detection method according to claim 2, characterized in that: Step 3) also includes the identification of mutually entangled fighting behaviors, which requires the following conditions to be met simultaneously: The velocities (v, t) of both bodies are less than the threshold v1; In the s seconds before and after the current time t, the accelerations of both bodies (a, t) are less than the threshold a1; The distance between individuals (d, t) is less than the threshold d0; The change per unit time of the movement direction (α, t) of the two bodies is greater than the threshold α1; Among them, v0>v1, a0>a1, α0<α1.

4. The detection method according to claim 3, characterized in that: In step 2), the individual continuous trajectories are calculated by spline interpolation or cubic polynomial interpolation.

Citation Information

Patent Citations

  • Intelligent golden-eye alarming method and device for identifying personnel gathering affray

    CN106127814A

  • Method for identifying fighting behavior in video monitoring system

    CN115661918A