CN115812612A - Multi-target mouse whole-growth-period high-throughput phenotype acquisition and analysis system and method - Google Patents

Abstract

The invention relates to a multi-target mouse whole-growth-period high-throughput phenotype acquisition and analysis system and method. According to the invention, a mouse cage system suitable for the life of the multi-target experimental mouse is constructed, the experimental mouse is observed from two rectangular holes on the top surface of the mouse cage by adopting two depth cameras to obtain RGB image data and depth image data, and meanwhile, the video data of the experimental mouse is obtained by adopting the RGB cameras to observe downwards from a round-corner rectangular hole on the top surface of the mouse cage. The method comprises the steps of identifying mice in real time and accurately tracking each mouse based on a deep learning network and a multi-target tracking algorithm, extracting key points such as the nose, ears, necks, central points of the back, tail roots and the like of each mouse, estimating behaviors of a single mouse and social behaviors among the mice according to the key points, determining behavior parameters such as time, duration, frequency and the like of occurrence of each behavior, and analyzing the health state of the mice according to the behavior parameters. The method can accurately extract the phenotype of the multi-target mouse in the whole life cycle, and is low in cost and convenient and quick to operate.

Description

Multi-target mouse full growth period high-throughput phenotype acquisition and analysis system and method

technical field

The invention belongs to the technical field of animal behavior and automated experimental equipment, and in particular relates to a system and method for acquiring and analyzing multi-target mouse high-throughput phenotypes throughout the growth period.

Background technique

Animal behavior is the body language of animals to express their psychology and physiology, and it is also the embodiment of animals' comprehensive functions. Animal ethology is a subject that studies the interaction between animals and the environment and other organisms. It not only observes the natural behavior of animals, but more importantly, observes the behavior and changes of animals under specific laboratory conditions, and then studies Experimental animal's nervous function, psychological process, analysis of drug effects, etc.

Mice are typical model animals. As mammals, their genes are more than 98% similar to human genes. Many diseases that are difficult to cure for humans can be simulated in mice. The analysis of different behaviors of mice is widely used It is used to measure the effect of experiments in fields such as biology, neuroscience, pharmacology and genetics.

The traditional measurement method of mouse behavior analysis is based on statistical analysis of the results of manual observation, which is time-consuming and laborious. Not only can it not realize some experiments that require long-term observation, but also cannot perform objective evaluation and measurement of behavior, and subjective factors have a great influence.

There are still some obvious deficiencies in the existing mouse behavior detection based on machine vision. First of all, the detection throughput is low, the number of target mice is small, and it is mainly suitable for short-term experiments. The equipment and methods for detecting the behavior of mice in the whole life cycle have not been reported, so the existing technology cannot observe the life cycle of different species of mice at the same time. phenotypes, resulting in deviations in the association analysis between gene-nutrition-phenotype of mice in later studies. Secondly, due to the extremely similar appearance of mice, the tracking accuracy of multi-target mice is not high, and it is easy to lose track and inaccurate tracking, which leads to the exchange of IDs between mice and data confusion. In addition, in the existing technology, RGB cameras are mostly used to obtain the two-dimensional phenotype of mice, and the extraction of three-dimensional mouse phenotypes and multi-target mouse tracking have not been reported yet. However, the depth information added to the three-dimensional point cloud can make use of the depth information more Accurately track multi-target mice, and phenotypes that cannot be obtained from 2D data can be obtained through 3D point clouds, such as mouse weight, mouse nesting behavior, etc.

Contents of the invention

(1) Technical problems to be solved

In order to solve technical problems such as low throughput, inaccurate tracking, short test period and incomplete acquisition of phenotypic parameters, the present invention proposes a system and method for high-throughput phenotype acquisition and analysis of multi-target mice throughout the growth period.

(2) Technical solutions

In order to solve the technical problem, the present invention provides a system and method for acquiring and analyzing multi-target high-throughput phenotypes in the whole growth period of mice.

A multi-target mouse full growth period high-throughput phenotype acquisition and analysis system is characterized in that it includes a mouse cage system, a shelf, a T-shaped bracket system, a depth camera, an RGB camera, a notebook computer, a switch, a video recorder, a display screens and workstations, where,

The mouse cage system includes multiple sets of mouse cage systems, each set of mouse cage systems includes mouse cages, drawers, water bottles and food boxes; the mouse cages are designed in a rectangular box, which is used to place experimental mice, which is convenient for experimental mice to live in the mouse cage free movement; the top surface of the mouse cage is provided with a plurality of small round holes for ventilation, and there are rectangular holes on both sides of the top surface; water bottles and food boxes are detachably installed on the side walls of the mouse cage, respectively It is used to provide water and food for experimental mice; the drawer is located directly under the mouse cage, and is covered with corncob bedding to prevent mouse excrement from polluting the mouse cage;

The shelf adopts a multi-layer shelf design, and each floor can place multiple mouse cage systems side by side at equal intervals; each layer of the shelf is equipped with multiple buckles, and each mouse cage system can be fixed to the shelf through the buckle. detachable installation;

The T-shaped bracket system includes multiple T-shaped brackets, which are fixedly installed on the shelf in an inverted T-shaped style, so that it is just above each mouse cage system;

The depth cameras are respectively installed at the ends of both sides of the lower end crossbar of each T-shaped bracket, and the depth cameras are directly facing the inside of the mouse cage through two rectangular holes on the top surface of the mouse cage to observe the experimental mice;

An RGB camera is detachably installed at the center of the top surface of each mouse cage. The RGB camera has infrared night vision function and can continuously observe the experimental mice for 24 hours;

The laptop is connected to the depth camera to store the RGB image data and depth image data collected by the depth camera;

The video recorder is connected to the RGB camera through a switch to store the video data collected by the RGB camera in real time;

The display screen is connected with the video recorder for real-time display and monitoring of the activities of the mice in all cages;

The workstation is connected with a laptop and a video recorder to process and analyze the video and image files of all mice acquired by the RGB camera and the depth camera, and obtain two-dimensional and three-dimensional phenotypic parameters of the whole life cycle of multi-target mice, including diet, drinking The frequency and duration of various behaviors including water, sleeping, standing, fighting and chasing were analyzed, and the health of the mice was analyzed according to the obtained result parameters.

Preferably, a rectangular hole with rounded corners is provided at the center of the top surface of the squirrel cage, and a small groove for passing cables is provided on one side of the rectangular hole with rounded corners; the RGB camera is detachably mounted on the rounded rectangular hole. The shape and size of the hole match the rounded rectangular plate, and the rounded rectangular plate is also provided with a groove corresponding to the rounded rectangular hole for passing the cable; the rounded rectangular plate is designed to be Snap into the rounded rectangular hole, so that the RGB camera is located in the inner space of the mouse cage and observes towards the mouse. The cable of the RGB camera is exported through a small groove and connected to an external device; The rounded rectangular board with the RGB camera installed is taken out of the rounded rectangular hole.

Preferably, the shelf is made of 304 stainless steel.

Preferably, the squirrel cage is composed of 6 transparent polyphenylene sulfone resin plates with a length, width and height of 265×265×310 mm.

Preferably, the two depth cameras above the squirrel cage are aimed at the inner space of the squirrel cage in an oblique manner, the inclination angle is 65° with the horizontal direction, and the height difference between the bottom of the depth camera and the top of the squirrel cage is 1cm.

A high-throughput phenotype acquisition and analysis method for the whole growth period of multi-target mice, which uses the aforementioned high-throughput phenotype acquisition and analysis system for the whole growth period of multi-target mice for operation, and specifically includes the following steps:

Step 1, the shelf and rat cage system are disinfected;

Step 2, install multiple squirrel cage systems on the shelf through buckles, and ensure that the two depth cameras on the T-shaped bracket are aligned with the inner space of the squirrel cage through the rectangular holes on both sides of the top surface of the squirrel cage;

Step 3, snap the rounded rectangular plate with the RGB camera into the rounded rectangular hole on the top surface of the squirrel cage, and ensure that the RGB camera is aligned with the inner space of the squirrel cage;

Step 4, cover the bottom of the drawer of the mouse cage with corncob bedding, and put the mouse into the mouse cage;

Step 5, the system is powered on, the depth camera acquires the RGB image and the depth image of the mouse, the RGB camera acquires the video data of the mouse, and the data is stored in real time;

Step 6. The staff regularly removes the mouse cage system, removes the mice from the mouse cage, and disinfects the mouse cage system, replaces the corncob bedding, reinstalls the mouse cage system and the rounded rectangular plate, and puts the mouse cage system into a small Rat and continue to observe;

Step 7, the workstation regularly performs digital image processing and analysis on the images and videos of the mice, completes the observation of the whole life cycle of the multi-objective mice, obtains the phenotypic parameter information of the mice, and analyzes the health of the mice based on the obtained result parameters situation.

Preferably, in step 7, the deep learning neural network YOLOv7 and the improved StrongSORT multi-target tracking algorithm are used to process and analyze the video data collected by the RGB camera to obtain the two-dimensional phenotypic parameter information of the mouse, the specific method is as follows:

(1) Annotate the video image, make a data set, and train the mouse recognition model based on YOLOv7 offline;

(2) Input mouse video data, use YOLOv7 mouse recognition model to identify multi-target mice in each frame of video and generate target detection frames, and assign a specific target detection frame to each target detection frame based on the improved StrongSORT multi-target tracking algorithm ID number and tracking;

(3) According to the detection results of YOLOv7 mice, identify multiple key points of each mouse, including the mouth, ears, neck, back center and tail root; and repair technology to repair key points;

(4) Estimate the posture of each mouse in each frame of video according to the key point position information;

(5) Calculate the behavior of each mouse and the social behavior between mice according to the posture and time series of mice in each frame of video images.

Preferably, in step 7, the specific improvement of the multi-target mouse tracking algorithm of the improved StrongSORT is as follows: in the current frame video, adopt YOLOv7 to identify the mouse and generate a detection frame, and combine the generated detection frame with the previous frame video After cascade matching and IOU matching are performed on the detection frame with a specific ID number, that is, the trajectory, if there is a trajectory that does not match the detection frame and a detection frame that does not match the detection frame, calculate the Euclidean distance between them, and minimize the Euclidean distance The ID number of the track that does not match the upper detection box is assigned to the detection box that does not match the upper track.

Preferably, in step 7, the RGB image and the depth image data collected by two depth cameras are used for analysis and processing, and the three-dimensional phenotype parameter information of the mouse is obtained, the specific method is as follows:

(1) Align the RGB image with the depth image, reconstruct a 3D point cloud through the depth image, assign the RGB value of the 3D point cloud, that is, color, through the aligned RGB image, and reconstruct the 3D point cloud with two depth cameras;

(2) Register the point clouds reconstructed by the two depth cameras through the external parameter calibration between the two depth cameras, and synthesize them into a 3D point cloud;

(3) Divide the reconstructed 3D point cloud into mouse and background, mark the data set, and train the pointnet++ network to extract the mouse point cloud from the background;

(4) Predict the next position of the mouse based on the Kalman filter, and use the Hungarian algorithm to match and track the predicted position with the mouse position segmented in (3);

(5) Segment the three-dimensional point cloud of the mouse into four parts: head, back, tail, and tail, mark the data set, and train the pointnet++ network;

(6) Estimate the posture of each mouse according to the mouse position information;

(7) Calculate the behavior of each mouse and the social behavior between mice according to the posture and time series of the mice in the point cloud.

Preferably, in step 7, the phenotypic parameter information of the mouse is obtained, including the frequency and duration of the behaviors of eating, drinking, sleeping, standing, fighting and chasing, and analyzing the health status of the mouse according to the obtained result parameters, Specifically, the following refinement steps are adopted:

(1) Identify and detect various behaviors of mice:

For eating and drinking, if the mouse mouth is located in the food box for more than 2 seconds, it is judged as eating, and the time when the mouse mouth is located in the food box is recorded as the duration;

To drink water, if the mouse's mouth is located at the mouth of the water bottle for more than 2 seconds, it is judged to be drinking water, and the time when the mouse's mouth is located at the mouth of the water bottle is recorded as the duration;

Sleeping, calculate the amount of change in the position of key parts of the mouse, including the head, back and tail, in adjacent frame images. When the change is less than 5 pixels, it is judged that the mouse is still. When the time of stillness is greater than 100 seconds , it is judged as sleeping, and the duration is recorded;

Standing, let the position of the mouse’s mouth in the image be A, the position of the center point of the back be B, the position of the root of the tail be C, the distance between the mouth and the center point of the back is a, the distance between the center point of the back and the root of the tail is b, If 0.9<a/b<1.1 and the angle between AB and BC is less than 30°, judge standing and record the duration;

Fighting, the relative movement speed of two mice is greater than 10cm/s, and the next frame is in contact with each other, then it is judged as a fight, and the duration is recorded;

Chasing, two mice moving in the same phase, one mouse’s mouth is behind the other mouse’s tail, and both speeds are greater than 5cm/s, and the relative speed of the two is less than 2cm/s, it is judged as chasing, and record the duration;

(2) Counting the occurrence times of various behaviors listed in 24 hours a day is the frequency, respectively accumulating the duration of each behavior, and recording the frequency and duration of each behavior;

(3) Establish a health index according to various behaviors and their frequency of occurrence and duration, and analyze the health status of the mice according to the health index.

(3) Beneficial effects

Compared with the prior art, the beneficial effects of the present invention are reflected in at least three aspects.

First of all, the mouse cage system designed and manufactured by the present invention meets the requirements of SPF level animal feeding. It uses an inverted T-shaped bracket to fix the depth camera, a rounded rectangular plate to fix the RGB camera, and a drawer to place padding, which is convenient for the installation of the hardware structure and sensor components. Disassembly makes it more convenient and quick to regularly disinfect the mouse cage and replace the litter, and provides a solid hardware guarantee for obtaining and analyzing the 2D and 3D phenotypes of multi-species and multi-target mice throughout the growth period.

Secondly, the StrongSORT multi-target tracking algorithm is improved. In the tracking process, after cascade matching and IOU matching, if there are trajectories that do not match the detection frame and detection frames that do not match the detection frame, a secondary match is performed to solve the problem. It solves the problems of blank frames and matching errors in the process of multi-target mouse tracking, effectively improves the accuracy of multi-target tracking, and ensures the quality of scientific research data related to mouse gene-nutrition-phenotype in the later stage.

In addition, the present invention can accurately extract the two-dimensional and three-dimensional phenotypes of multiple mice in multiple mouse cages at the same time, and is suitable for both short-term experiments and full-life cycle experiments of mice. RGB camera and two depth cameras are used to accurately obtain two-dimensional and three-dimensional phenotypes, and the acquisition of phenotypic parameters is more comprehensive, and the health status of mice can be evaluated according to the two-dimensional and three-dimensional phenotypes of mice. The association analysis between nutrition and phenotype is of great significance.

Finally, the structure design of the squirrel cage system provided by the present invention is modular, the cost is relatively small, the operation is convenient and fast, and it has the potential for large-scale popularization and application.

Description of drawings

Fig. 1 is a schematic structural view of the shelf and mouse cage system of the present invention.

Fig. 2 is a structural schematic diagram (head-up view) of the mouse cage system and the sensor assembly.

Fig. 3 is a structural schematic diagram (slanted top view) of the mouse cage system and sensor components.

Fig. 4 is the working flowchart embodiment of the present invention.

Figure 5 shows the results of multi-target mouse tracking based on the original StrongSORT.

Figure 6 shows the results of multi-target mouse tracking based on the improved StrongSORT.

Detailed ways

In order to solve the technical problem, the present invention provides a system and method for acquiring and analyzing multi-target high-throughput phenotypes in the whole growth period of mice. The technical solution of the present invention will be further explained in conjunction with the accompanying drawings.

The present invention provides a system and method for obtaining and analyzing high-throughput phenotypes of multi-target mice throughout the growth period. The system hardware design is shown in Figure 1, Figure 2 and Figure 3, including a mouse cage system, a shelf, and a T-shaped Mounting systems, depth cameras, RGB cameras, laptops, switches, recorders, displays and workstations.

The squirrel cage system includes multiple sets of squirrel cage systems, and each set of squirrel cage system includes squirrel cages, drawers, water bottles and food boxes; the squirrel cages are custom-made and adopt a rectangular box design, which is composed of six transparent PPSU boards through 304 stainless steel M4 Screw fixed, PPSU material and 304 stainless steel are anti-corrosion, high temperature resistant and durable. The mouse cage is used to house the experimental mice, which is convenient for the experimental mice to move freely inside the mouse cage; there are 26 circles with a diameter of 15mm on the top surface of the mouse cage. Holes, a rounded rectangular hole with a length x width of 96 x 56mm and two rectangle holes with a length x width of 110 x 40mm, of which 26 holes with a diameter of 15mm are used to diffuse the odor in the mouse cage; water bottles and food boxes It is detachably installed on the side wall of the mouse cage, and the height from the bottom of the mouse cage is 150mm and 50mm respectively, which are used to provide water and food for the experimental mice respectively; the drawer is located directly under the mouse cage, and the corn cob pad is used It is fully paved to prevent mouse excrement from polluting the cage, and the drawer structure makes it more convenient and quick to replace the bedding and manually weigh the mice.

The shelf adopts a multi-layer shelf design, and each floor can place multiple mouse cage systems side by side at equal intervals; each layer of the shelf is equipped with multiple buckles, and each mouse cage system can be fixed to the shelf through the buckle. The detachable installation is convenient for the regular disinfection of the mouse cage and the position is fixed after each disassembly and installation, so that the field of view of the RGB camera and the depth camera will not change when collecting data.

The T-Bracket System consists of multiple T-Brackets fixedly mounted on a shelf in an inverted-T pattern so that it sits just above each mouse cage system.

Two depth cameras are respectively installed at the two ends of the lower crossbar of each T-shaped bracket, and the two depth cameras are oblique to monitor the experimental mice in the cage through the two rectangular holes on the top surface of the cage. Observation, the inclination angle is 65° from the horizontal direction, the height difference between the bottom of the depth camera and the top of the squirrel cage is 1cm, and the two depth cameras are connected by an audio cable to ensure that the time stamps of the two depth cameras are consistent when collecting data , the laptop is connected to the depth camera through a USB cable, and is used to store the RGB image data and depth image data collected by the depth camera.

The rounded rectangular hole on the top surface of the squirrel cage is provided with a small groove on one side for passing cables; the RGB camera is detachably mounted on a rounded rectangular plate matching the shape and size of the rounded rectangular hole, And the rounded rectangular plate is also provided with a small groove corresponding to the rounded rectangular hole for passing the cable; the rounded rectangular plate is designed to be snapped into the rounded rectangular hole, so that the RGB camera is located on the mouse Observe the inner space of the cage and face the mouse. The cable of the RGB camera is exported through a small groove to connect to the external device; take out.

The RGB camera has an infrared night vision function, which can continuously observe the experimental mice for 24 hours. The video recorder is connected to the RGB camera through a switch to store the video data collected by the RGB camera in real time.

The display screen is connected with a video recorder for real-time display and monitoring of the activities of mice in all cages.

The workstation is connected with a laptop and a video recorder to process and analyze the video and image files of all mice acquired by the RGB camera and the depth camera, and obtain two-dimensional and three-dimensional phenotypic parameters of the whole life cycle of multi-target mice, including diet, drinking The frequency and duration of various behaviors including water, sleeping, standing, fighting and chasing were analyzed, and the health of the mice was analyzed according to the obtained result parameters.

A method for obtaining and analyzing a multi-target mouse high-throughput phenotype throughout the growth period, the workflow diagram is shown in Figure 4, and the steps are as follows: (1) put a mouse cage, an RGB camera, two depth cameras, Laptops, switches, video recorders, and monitors are connected. (2) Cover the bottom of the squirrel cage drawer with corncob bedding. (3) Put 3 mice on the litter, and put the drawer containing the mice into the mouse cage. (4) Start and power on, and the notebook computer and video recorder collect image data and video data respectively. (5) The workstation regularly performs digital image processing and analysis on the image data and video data to obtain the two-dimensional and three-dimensional phenotypes of the mice, including activity paths, behavioral states, and the frequency and duration of various behaviors.

The deep learning neural network YOLOv7 and the improved StrongSORT multi-target tracking algorithm are used to process and analyze the video data collected by the RGB camera, and obtain the two-dimensional phenotypic parameter information of the mouse, which specifically includes the following aspects: (1) Annotate the video image and make a data set , offline training based on YOLOv7 mouse recognition model; (2) Modify the multi-target mouse tracking algorithm based on StrongSORT, and combine the detection frame generated by YOLOv7 in the current frame with the detection frame assigned a specific ID number in the previous frame video, that is, the trajectory After cascade matching and IOU matching, if there are tracks that do not match the upper detection frame and detection frames that do not match the upper track, calculate the Euclidean distance between them, and set the ID of the track that does not match the upper detection frame with the smallest Euclidean distance Numbers are assigned to detection boxes that do not match the upper trajectory. Figure 5 is the multi-target tracking result based on the original StrongSORT. The left picture shows the tracking result when the mouse climbed the barbed wire, and the right picture shows the tracking result when the mouse jumped from the barbed wire to the ground. It can be seen from the figure that the mouse climbed the barbed wire and ID number changed from 3 to 4 when jumping to the ground. Figure 6 shows the multi-target tracking results based on the improved StrongSORT. The left picture shows the mouse climbing the barbed wire, and the right picture shows the mouse jumping to the ground. It can be seen from the figure that the ID number of the mouse does not change when it climbs the barbed wire and jumps to the ground. , that is, the tracking ID number of the same mouse is accurate; (3) According to the detection results of YOLOv7 mice, identify key points including the mouth, ears, neck, back center, and tail root of each mouse. If there is occlusion, Then repair the key points according to the key point information and repair technology of several frames before and after; (4) estimate the posture of each mouse in each frame of video image according to the key point position information; (5) according to the posture and time of the mouse in each frame of video image The sequence calculates the behavior of each mouse and the social behavior between mice.

The RGB image and depth image data collected by two depth cameras are used for analysis and processing, and the three-dimensional phenotypic parameters of the mouse are obtained, including the following aspects: (1) Align the RGB image with the depth image, and reconstruct it through the depth image The three-dimensional point cloud is generated, and the RGB value of the three-dimensional point cloud is assigned to the RGB image after the alignment, and the two depth cameras reconstruct the three-dimensional point cloud; (2) the two depth cameras are reconstructed through the external parameter calibration between the two depth cameras (3) Divide the reconstructed 3D point cloud into mouse and background, mark the data set, and train the pointnet++ network to extract the mouse point cloud from the background; (4) Predict the next position of the mouse based on the Kalman filter, and use the Hungarian algorithm to match and track the predicted position with the mouse position segmented in (3); (5) Segment the three-dimensional point cloud of the mouse into head , back, tail, and four parts of the tail, mark the data set, and train the pointnet++ network; (6) estimate the posture of each mouse according to the location information of the mouse; (7) calculate according to the posture and time series of the mouse in the point cloud Behavior of each mouse and social behavior among mice.

The specific embodiments described in this application are merely illustrative of the spirit of the invention. Those skilled in the art to which the present invention belongs may make various modifications or supplements to the specific embodiments or replace them in similar ways, but they will not deviate from the spirit of the present invention or go beyond the scope defined in the appended claims.

Claims (10)

1. A multi-target mouse full growth period high-throughput phenotype acquisition and analysis system, characterized in that it includes a mouse cage system, a shelf, a T-shaped bracket system, a depth camera, an RGB camera, a notebook computer, a switch, a video recorder, a display screens and workstations, where, The mouse cage system includes multiple sets of mouse cage systems, each set of mouse cage systems includes mouse cages, drawers, water bottles and food boxes; the mouse cages are designed in a rectangular box, which is used to place experimental mice, which is convenient for experimental mice to live in the mouse cage free movement; the top surface of the mouse cage is provided with a plurality of small round holes for ventilation, and there are rectangular holes on both sides of the top surface; water bottles and food boxes are detachably installed on the side walls of the mouse cage, respectively It is used to provide water and food for experimental mice; the drawer is located directly under the mouse cage, and is covered with corncob bedding to prevent mouse excrement from polluting the mouse cage; The shelf adopts a multi-layer shelf design, and each floor can place multiple mouse cage systems side by side at equal intervals; each layer of the shelf is equipped with multiple buckles, and each mouse cage system can be fixed to the shelf through the buckle. detachable installation; The T-shaped bracket system includes multiple T-shaped brackets, which are fixedly installed on the shelf in an inverted T-shaped style, so that it is just above each mouse cage system; The depth cameras are respectively installed at the ends of both sides of the lower end crossbar of each T-shaped bracket, and the depth cameras are directly facing the inside of the mouse cage through two rectangular holes on the top surface of the mouse cage to observe the experimental mice; An RGB camera is detachably installed at the center of the top surface of each mouse cage. The RGB camera has infrared night vision function and can continuously observe the experimental mice for 24 hours; The laptop is connected to the depth camera to store the RGB image data and depth image data collected by the depth camera; The video recorder is connected to the RGB camera through a switch to store the video data collected by the RGB camera in real time; The display screen is connected with the video recorder for real-time display and monitoring of the activities of the mice in all cages; The workstation is connected with a laptop and a video recorder to process and analyze the video and image files of all mice acquired by the RGB camera and the depth camera, and obtain two-dimensional and three-dimensional phenotypic parameters of the whole life cycle of multi-target mice, including diet, drinking The frequency and duration of various behaviors including water, sleeping, standing, fighting and chasing were analyzed, and the health of the mice was analyzed according to the obtained result parameters. 2. multi-objective mouse whole growth period high-throughput phenotype acquisition and analysis system according to claim 1, is characterized in that, the center of the top surface of the mouse cage is provided with a rounded rectangular hole, and the rounded rectangular hole is in a The side is provided with a small groove for passing the cable; the RGB camera is detachably installed on the rounded rectangular board matching the shape and size of the rounded rectangular hole, and the rounded rectangular board is also provided with The groove corresponding to the rounded rectangular hole is used to pass the cable; the rounded rectangular plate is designed to be snapped into the rounded rectangular hole, so that the RGB camera is located in the inner space of the mouse cage and carried out towards the mouse. For observation, the cable of the RGB camera is exported through a small groove and connected to external equipment; after the observation is completed, the rounded rectangular plate with the RGB camera installed can be taken out of the rounded rectangular hole by lifting the cable. 3. The multi-target mouse full growth period high-throughput phenotype acquisition and analysis system according to claim 1, wherein the shelf is made of 304 stainless steel. 4. The multi-objective mouse full growth period high-throughput phenotype acquisition and analysis system according to claim 1, wherein the mouse cage is composed of 6 transparent polyphenylene sulfone resin plates, and the length and width The height is 265×265×310mm. 5. The multi-target mouse full growth period high-throughput phenotype acquisition and analysis system according to claim 1, characterized in that, the two depth cameras above the mouse cage are all aimed at the inner space of the mouse cage in an oblique manner, and the oblique The angles are all 65° with the horizontal direction, and the height difference between the bottom of the depth camera and the top of the mouse cage is 1cm. 6. The method for obtaining and analyzing high-throughput phenotypes of the whole growth period of multi-target mice, which adopts the high-throughput phenotype acquisition and analysis system for the whole growth period of multi-target mice as claimed in any one of claims 1-5 Carry out the work, including the following steps: Step 1, the shelf and rat cage system are disinfected; Step 2, install multiple squirrel cage systems on the shelf through buckles, and ensure that the two depth cameras on the T-shaped bracket are aligned with the inner space of the squirrel cage through the rectangular holes on both sides of the top surface of the squirrel cage; Step 3, snap the rounded rectangular plate with the RGB camera into the rounded rectangular hole on the top surface of the squirrel cage, and ensure that the RGB camera is aligned with the inner space of the squirrel cage; Step 4, cover the bottom of the drawer of the mouse cage with corncob bedding, and put the mouse into the mouse cage; Step 5, the system is powered on, the depth camera acquires the RGB image and the depth image of the mouse, the RGB camera acquires the video data of the mouse, and the data is stored in real time; Step 6. The staff regularly removes the mouse cage system, removes the mice from the mouse cage, and disinfects the mouse cage system, replaces the corncob bedding, reinstalls the mouse cage system and the rounded rectangular plate, and puts the mouse cage system into a small Rat and continue to observe; Step 7, the workstation regularly performs digital image processing and analysis on the images and videos of the mice, completes the observation of the whole life cycle of the multi-objective mice, obtains the phenotypic parameter information of the mice, and analyzes the health of the mice based on the obtained result parameters situation. 7. The method for obtaining and analyzing high-throughput phenotypes of multi-target mice throughout the growth period according to claim 6, characterized in that: in step 7, the deep learning neural network YOLOv7 and the improved StrongSORT multi-target tracking algorithm are used for processing and analysis The video data collected by the RGB camera is used to obtain the two-dimensional phenotypic parameter information of the mouse. The specific method is as follows: (1) Annotate the video image, make a data set, and train the mouse recognition model based on YOLOv7 offline; (2) Input mouse video data, use YOLOv7 mouse recognition model to identify multi-target mice in each frame of video and generate target detection frames, and assign a specific target detection frame to each target detection frame based on the improved StrongSORT multi-target tracking algorithm ID number and tracking; (3) According to the detection results of YOLOv7 mice, identify multiple key points of each mouse, including the mouth, ears, neck, back center and tail root; and repair technology to repair key points; (4) Estimate the posture of each mouse in each frame of video according to the key point position information; (5) Calculate the behavior of each mouse and the social behavior between mice according to the posture and time series of mice in each frame of video images. 8. The method for obtaining and analyzing high-throughput phenotypes of multi-target mice throughout the growth period according to claim 7, characterized in that: in step 7, the specific improvement of the multi-target mouse tracking algorithm of the improved StrongSORT is as follows : In the current frame of video, YOLOv7 is used to identify the mouse and generate a detection frame. After cascading matching and IOU matching between the generated detection frame and the detection frame with a specific ID number in the previous frame of video, that is, the trajectory, if there is no match The track of the upper detection frame and the detection frame that does not match the upper track calculate the Euclidean distance between them, and assign the ID number of the track that does not match the upper detection frame with the smallest Euclidean distance to the detection frame that does not match the upper track. 9. The multi-target mouse full growth period high-throughput phenotype acquisition and analysis method according to claim 6, characterized in that: in step 7, the RGB images and depth image data collected by two depth cameras are used for analysis and processing , to obtain the three-dimensional phenotypic parameter information of the mouse, the specific method is as follows: (1) Align the RGB image with the depth image, reconstruct a 3D point cloud through the depth image, assign the RGB value of the 3D point cloud, that is, color, through the aligned RGB image, and reconstruct the 3D point cloud with two depth cameras; (2) Register the point clouds reconstructed by the two depth cameras through the external parameter calibration between the two depth cameras, and synthesize them into a 3D point cloud; (3) Divide the reconstructed 3D point cloud into mouse and background, mark the data set, and train the pointnet++ network to extract the mouse point cloud from the background; (4) Predict the next position of the mouse based on the Kalman filter, and use the Hungarian algorithm to match and track the predicted position with the mouse position segmented in (3); (5) Segment the three-dimensional point cloud of the mouse into four parts: head, back, tail, and tail, mark the data set, and train the pointnet++ network; (6) Estimate the posture of each mouse according to the mouse position information; (7) Calculate the behavior of each mouse and the social behavior between mice according to the posture and time series of the mice in the point cloud. 10. The multi-target mouse full growth period high-throughput phenotype acquisition and analysis method according to claim 6, characterized in that: in step 7, the phenotypic parameter information of the mouse includes diet, drinking water, sleep, The frequency and duration of the behaviors of standing, fighting and chasing, and analyze the health status of the mice according to the obtained result parameters, specifically adopt the following refinement steps: (1) Identify and detect various behaviors of mice: For eating and drinking, if the mouse mouth is located in the food box for more than 2 seconds, it is judged as eating, and the time when the mouse mouth is located in the food box is recorded as the duration; To drink water, if the mouse's mouth is located at the mouth of the water bottle for more than 2 seconds, it is judged to be drinking water, and the time when the mouse's mouth is located at the mouth of the water bottle is recorded as the duration; Sleeping, calculate the amount of change in the position of key parts of the mouse, including the head, back and tail, in adjacent frame images. When the change is less than 5 pixels, it is judged that the mouse is still. When the time of stillness is greater than 100 seconds , it is judged as sleeping, and the duration is recorded; Standing, let the position of the mouse’s mouth in the image be A, the position of the center point of the back be B, the position of the root of the tail be C, the distance between the mouth and the center point of the back is a, the distance between the center point of the back and the root of the tail is b, If 0.9<a/b<1.1 and the angle between AB and BC is less than 30°, judge standing and record the duration; Fighting, the relative movement speed of two mice is greater than 10cm/s, and the next frame is in contact with each other, then it is judged as a fight, and the duration is recorded; Chasing, two mice moving in the same phase, one mouse’s mouth is behind the other mouse’s tail, and both speeds are greater than 5cm/s, and the relative speed of the two is less than 2cm/s, it is judged as chasing, and record the duration; (2) Counting the occurrence times of various behaviors listed in 24 hours a day is the frequency, respectively accumulating the duration of each behavior, and recording the frequency and duration of each behavior; (3) Establish a health index according to various behaviors and their frequency of occurrence and duration, and analyze the health status of the mice according to the health index.

Images