CN-121970695-A - Animal social obstacle behavior detection device and method, electronic equipment and medium

CN121970695ACN 121970695 ACN121970695 ACN 121970695ACN-121970695-A

Abstract

The application relates to the technical field of animal behaviours, in particular to an animal social obstacle behavior detection device and method, electronic equipment and medium. The animal social obstacle behavior detection device provided by the application consists of an observation box body, an image acquisition device and a data processing unit. The interior of the observation box body is divided into a first subchamber and a second subchamber which are respectively used for placing experimental animals and control animals, so that behavior observation of the two groups of animals under the same environment is ensured. The multi-view image acquisition device in each subchamber captures animal activities from different angles through a plurality of cameras to generate multi-view moving pictures. The data processing unit is responsible for extracting animal behavior tracks from the pictures, and analyzing the tracks by utilizing a pre-trained phenotype classification model so as to detect social barrier behaviors, thereby realizing efficient and quantitative analysis of animal behaviors.

Inventors

WU SHENGXI
CAI GUOHONG
WANG WENTING
WAN XIANGDONG
YANG DINGDING

Assignees

中国人民解放军空军军医大学

Dates

Publication Date: 20260505
Application Date: 20251230

Claims (10)

1. An animal social barrier behavior detection device, comprising: the observation box body is internally divided into a first subchamber and a second subchamber which are mutually isolated, wherein the first subchamber is used for accommodating experimental animals, the second subchamber is used for synchronously accommodating control animals under the same environmental condition of the first subchamber, and the experimental animals and the control animals are the same kind of animals which make different behavior expressions; the image acquisition device comprises a plurality of cameras arranged from different shooting visual angles, and is used for shooting the behavior activities of animals from a plurality of shooting visual angles, wherein each camera corresponds to one shooting visual angle; The image acquisition device arranged in the first subchamber is a first multi-view image acquisition device, and the first multi-view image acquisition device is used for acquiring experimental multi-view moving pictures of the experimental animal; the image acquisition device arranged in the second subchamber is a second multi-view image acquisition device, and the second multi-view image acquisition device is used for acquiring a comparison multi-view moving picture of the comparison animal; The data processing unit is connected with the first multi-view image acquisition device and the second multi-view image acquisition device, and is used for extracting experimental animal behavior tracks from the experimental multi-view moving pictures, extracting comparison animal behavior tracks from the comparison multi-view moving pictures, and carrying out social obstacle behavior detection on the experimental animal behavior tracks and the comparison animal behavior tracks through a pre-trained phenotype classification model to obtain animal social obstacle behavior detection data.
2. A method for detecting social barrier behavior of an animal, applied to the apparatus for detecting social barrier behavior of an animal according to claim 1, the method comprising: the method comprises the steps that multi-view image acquisition is carried out on an experimental animal accommodated in a first subchamber through a first multi-view image acquisition device, so that an experimental multi-view moving picture of the experimental animal is obtained; performing multi-view image acquisition on a control animal accommodated in a second subchamber through a second multi-view image acquisition device to obtain a control multi-view moving picture of the control animal; Extracting an experimental animal behavior track from the experimental multi-view moving picture; Extracting a control animal behavior track from the control multi-view moving picture; And carrying out social obstacle behavior detection on the experimental animal behavior track and the control animal behavior track through a pre-trained phenotype classification model to obtain animal social obstacle behavior detection data.
3. The method according to claim 2, further comprising pre-training the phenotype classification model before the social barrier behavior detection is performed on the experimental animal behavior trace and the control animal behavior trace by the pre-trained phenotype classification model to obtain animal social barrier behavior detection data, specifically comprising: determining an original classification detection model; the method comprises the steps of obtaining a behavior detection training data set, wherein the behavior detection training data set comprises a plurality of behavior phenotype sample tracks and behavior phenotype marking data corresponding to each behavior phenotype sample track; and performing supervision training on the classification detection model based on each behavior phenotype sample track and the behavior phenotype labeling data corresponding to each behavior phenotype sample track to obtain the pre-trained phenotype classification model.
4. A method according to claim 3, wherein said acquiring a behavior detection training dataset comprises: obtaining a plurality of said behavioral phenotype sample trajectories, wherein each of said behavioral phenotype sample trajectories corresponds to a sample behavioral phenotype of a sample animal; performing two-dimensional body tracking on each behavior phenotype sample track to obtain a plurality of sample body tracking data; reconstructing three-dimensional behaviors based on a plurality of sample body tracking data of each behavior phenotype sample track to obtain three-dimensional behavior characterization data; performing behavior clustering on each three-dimensional behavior characterization data to obtain a plurality of behavior data clustering clusters; Labeling each behavior data clustering cluster to obtain behavior phenotype labeling data corresponding to each behavior phenotype sample track.
5. A method according to claim 3, wherein said performing supervised training of said classification detection model based on each of said behavioral phenotype sample trajectories and said behavioral phenotype labeling data corresponding to each of said behavioral phenotype sample trajectories to obtain a pre-trained said phenotype classification model comprises: Inputting the behavior phenotype sample track into the classification detection model for classification training to obtain a classification training result of the iterative round; Comparing the behavior phenotype labeling data corresponding to the behavior phenotype sample track with the classification training result to obtain training deviation data of the round; Adjusting model parameters of the classification detection model based on the training deviation data of the present round to update the classification detection model; And returning to input the behavior phenotype sample track into the updated classification detection model to obtain the classification training result of the next iteration round until the classification detection model meets the preset training expected condition, and obtaining the pre-trained phenotype classification model.
6. A method according to claim 3, wherein the social barrier behavior detection is performed on the experimental animal behavior trace and the control animal behavior trace by a pre-trained phenotype classification model to obtain animal social barrier behavior detection data, including performing social barrier behavior detection on the experimental animal behavior trace, specifically including: performing cluster analysis in a plurality of behavior data cluster clusters aiming at the behavior track of the experimental animal by using the phenotype classification model so as to determine a target behavior cluster; And determining the animal social obstacle behavior detection data according to the behavior phenotype marking data corresponding to the target behavior cluster.
7. The method according to claim 2, wherein the social barrier behavior detection of the experimental animal behavior trace and the control animal behavior trace by the pre-trained phenotype classification model to obtain animal social barrier behavior detection data comprises: inputting the behavior track of the experimental animal into the phenotype classification model for carrying out phenotype identification to obtain the experimental behavior phenotype of the experimental animal; inputting the behavior track of the control animal into the phenotype classification model for carrying out phenotype identification to obtain the control behavior phenotype of the control animal; Performing difference calculation on the behavior track of the experimental animal and the behavior track of the control animal through the phenotype classification model to obtain an experimental quantitative difference index of the experimental animal relative to the control animal; Generating animal social barrier behavioral detection data based on the experimental behavioral phenotype, the control behavioral phenotype, and the experimental quantitative differential indicator.
8. The method of claim 2, wherein the extracting the experimental animal behavior trace from the experimental multi-view moving picture comprises: Based on the experimental multi-view moving pictures corresponding to each shooting view, tracking a two-dimensional body track of the experimental animal to obtain a two-dimensional behavior track; Acquiring camera calibration parameters corresponding to each shooting visual angle; carrying out three-dimensional modeling processing on a plurality of groups of camera calibration parameters corresponding to each other one by one and the two-dimensional behavior track to obtain a three-dimensional behavior track; And identifying the three-dimensional behavior track through a pre-trained animal gesture estimation model to obtain the experimental animal behavior track.
9. An electronic device comprising a memory, a processor, the memory storing a computer program, the processor implementing the method for detecting social barrier behavior of an animal according to any one of claims 2 to 8 when executing the computer program.
10. A computer-readable storage medium, characterized in that the storage medium stores a program that is executed by a processor to implement the detection method for animal social barrier behavior according to any one of claims 2 to 8.

Description

Animal social obstacle behavior detection device and method, electronic equipment and medium Technical Field The application relates to the technical field of animal behaviours, in particular to an animal social obstacle behavior detection device and method, electronic equipment and medium. Background The traditional detection method for animal social obstacle behaviors mainly relies on manual observation or two-dimensional video analysis, typical paradigms comprise an open field test, a three-box social experiment, a social interaction test, an object recognition test and the like, and the means are used for semi-quantitatively evaluating the social evasiveness, new and different environment exploration capability and the engraving repeated behaviors of a mouse by recording the position track of the animal in a specific scene, the contact frequency and the duration time of the animal and a target object or a companion. However, the method has limitation in data acquisition dimension and behavior decoding depth, plane track tracking under a single camera view can only be realized, coarse granularity space information of 'where the animal moves' is obtained, and fine action semantics of 'what behavior is in progress' cannot be accurately identified. Particularly in the aspect of quantification of the notch plate behaviors, due to the lack of continuous capture of three-dimensional motions of key parts of the body, the starting moment, duration and evolution mode of abnormal repeated motions are difficult to objectively judge, and specific behavior subtypes such as wool arrangement, jumping and rotation cannot be distinguished, so that a great amount of subjective judgment deviation is mixed in evaluation results, and the consistency coefficient among different experimenters is generally lower than the reference requirement of repeatability research. The one-sided performance makes it difficult for the detection data to comprehensively reflect the core phenotype characteristics of the autism model mice, and reduces the effectiveness of animal social disorder behavior detection. Therefore, how to efficiently detect the behavioral activities of animals to better realize quantitative analysis of the behavioral activities of animals has become a major problem to be solved in the industry. Disclosure of Invention The present application aims to solve at least one of the technical problems existing in the prior art. Therefore, the application provides the animal social obstacle behavior detection device and method, the electronic equipment and the medium, which can efficiently detect the behavior activity of the animal so as to better realize quantitative analysis of the animal behavior activity. An animal social barrier behavior detection apparatus according to an embodiment of the first aspect of the present application includes: the observation box body is internally divided into a first subchamber and a second subchamber which are mutually isolated, wherein the first subchamber is used for accommodating experimental animals, the second subchamber is used for synchronously accommodating control animals under the same environmental condition of the first subchamber, and the experimental animals and the control animals are the same kind of animals which make different behavior expressions; the image acquisition device comprises a plurality of cameras arranged from different shooting visual angles, and is used for shooting the behavior activities of animals from a plurality of shooting visual angles, wherein each camera corresponds to one shooting visual angle; The image acquisition device arranged in the first subchamber is a first multi-view image acquisition device, and the first multi-view image acquisition device is used for acquiring experimental multi-view moving pictures of the experimental animal; the image acquisition device arranged in the second subchamber is a second multi-view image acquisition device, and the second multi-view image acquisition device is used for acquiring a comparison multi-view moving picture of the comparison animal; The data processing unit is connected with the first multi-view image acquisition device and the second multi-view image acquisition device, and is used for extracting experimental animal behavior tracks from the experimental multi-view moving pictures, extracting comparison animal behavior tracks from the comparison multi-view moving pictures, and carrying out social obstacle behavior detection on the experimental animal behavior tracks and the comparison animal behavior tracks through a pre-trained phenotype classification model to obtain animal social obstacle behavior detection data. According to a second aspect of the present application, a method for detecting social obstacle behavior of an animal is applied to an apparatus for detecting social obstacle behavior of an animal according to the first aspect of the present application, and the method includes: the m