Animal information processing method and apparatus, and network device
By acquiring video data of animals throughout their entire life cycle, and combining visual understanding technology with multimodal large models, an animal information recognition model is trained. This solves the problems of incomplete information acquisition and low accuracy in existing technologies, and enables efficient acquisition of basic animal information, behavioral characteristics, and disease models.
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- SHENZHEN INST OF ADVANCED TECH CHINESE ACAD OF SCI
- Filing Date
- 2024-11-07
- Publication Date
- 2026-05-15
AI Technical Summary
Existing animal behavior analysis systems mainly rely on video analysis, which cannot obtain basic information about animals, behavioral characteristics, and disease models. Furthermore, the learning cost for researchers is high, the processing is complex, and the accuracy of the information needs to be improved.
By acquiring video data of animals throughout their entire life cycle, and combining visual understanding technology with multimodal large models, an animal information recognition model is trained to obtain basic information, behavioral characteristics, disease models, and biological indicators. This information is obtained directly from the video, reducing the cost of manual annotation and improving accuracy.
It enables the direct acquisition of basic, behavioral, and disease information of animals through video data, reducing researchers' processing time and improving the richness and accuracy of information acquisition. It is applicable to various experimental animal scenarios.
Smart Images
Figure CN2024130587_15052026_PF_FP_ABST
Abstract
Description
Animal information processing methods, devices and network equipment Technical Field
[0001] This application relates to the field of computer information technology, and more specifically, to an animal information processing method, apparatus, and network device. Background Technology
[0002] With the development of pose estimation and deep learning, animal behavior analysis systems have emerged and developed in recent years, mainly by extracting features from videos for behavior analysis.
[0003] However, existing solutions typically only analyze animal behavior based on video footage and cannot obtain other information.
[0004] Summary of the Invention
[0005] The embodiments of this application provide an animal information processing method, apparatus, and network device that can acquire video data of animals throughout their entire life cycle and obtain corresponding basic information, behavioral characteristics, disease models, biological indicators, etc. as annotations for the video data to train a recognition model. The trained recognition model can then be used to directly obtain the animal's basic information, behavioral characteristics, disease models, and biological indicators from the video.
[0006] The technical solution is as follows:
[0007] In a first aspect, this application provides an animal information processing method, the method comprising: acquiring first video data of an animal throughout its entire life cycle, the first video data corresponding to an animal's disease model, the video data including a first video corresponding to spontaneous animal behavior and a second video corresponding to an animal behavior test; acquiring first labeled data based on the first video data to train an animal information recognition model; the first labeled data including: basic information, behavioral characteristics, disease model, and biometrics; acquiring test video data and inputting it into the animal information recognition model for recognition to obtain test information recognition results, the test information recognition results including basic information, behavioral characteristics, disease model, and biometrics; analyzing the behavioral characteristics in the test information recognition results to filter action sequences with an accuracy rate lower than a preset threshold, and after correction, training the animal information recognition model to determine a well-trained animal information recognition model.
[0008] Furthermore, before acquiring video annotation data, the method further includes: inputting video data into a behavior analysis model, determining the behavior analysis results, and removing low-quality videos from the video data based on the behavior analysis results; the method further includes: acquiring video data to be analyzed and inputting it into a trained animal information recognition model to determine the animal's analysis results, wherein the analysis results include basic information, behavioral characteristics, disease models, and biometrics.
[0009] Furthermore, the steps for training the animal information recognition model include: acquiring second video data of animals and corresponding annotations, wherein the second video data includes videos of disease-free animals, a third video corresponding to spontaneous animal behavior, and a fourth video corresponding to animal behavior tests; acquiring the third video and its corresponding annotations, and training the animal information recognition model; acquiring the fourth video and its corresponding annotations, and training the animal information recognition model trained on the third video; acquiring the first video and its corresponding annotations, and training the animal information recognition model trained on the fourth video; acquiring the second video and its corresponding annotations, and training the animal information recognition model trained on the first video.
[0010] Furthermore, the steps for training the animal information recognition model include: obtaining the fifth video and its corresponding annotation, the annotation of which is obtained by analyzing the fifth video; obtaining the sixth video and its corresponding annotation, the annotation of which is obtained by collecting data from the animal using an auxiliary acquisition device; training the animal information recognition model based on the fifth video and its annotation, and training the animal information recognition model trained on the fifth video based on the sixth video and its annotation.
[0011] Furthermore, the basic information includes animal species, sex, age, and weight; the behavioral characteristics include behavioral definition, duration, and behavioral parameters; the disease model includes natural animals or disease model animals, disease course, and main manifestations; and the bioindicators include heart rate, blood pressure, respiratory rate, and electroencephalogram (EEG).
[0012] Furthermore, in the first video data of the animal's entire life cycle: the first video recording duration is 1 hour, and the second video recording duration is 5-15 minutes; the video collection frequency for animals with a lifespan of less than 3 years is once a week; the video collection frequency for animals with a lifespan of more than 3 years is once a month; the video recording is divided into single-view and multi-view; the multi-view consists of 4 peripheral cameras and 1 top camera; the single-view includes the top camera; basic information is collected when acquiring the first video data, and the first annotation data of the first video data is formed based on the basic information.
[0013] Furthermore, the first labeled data also includes sound information and ambient temperature information.
[0014] Furthermore, the step of analyzing the behavioral features in the test information recognition results to filter action sequences with an accuracy rate lower than a preset threshold, and then training the animal information recognition model after correction to determine the trained animal information recognition model, includes: analyzing the behavioral features in the test information recognition results by manual screening to filter action sequences with an accuracy rate lower than a preset threshold; filtering and correcting the selected action sequences and performing manual correction to determine test update data; and adjusting the animal information recognition model based on the test update data to determine the trained animal information recognition model.
[0015] Secondly, this application provides an animal information processing device, the device comprising: a video data acquisition module, used to acquire first video data of the animal's entire life cycle, the first video data corresponding to the animal's disease model, the video data including a first video corresponding to the animal's spontaneous behavior and a second video corresponding to the animal's behavior test; a recognition model training module, used to acquire first labeled data based on the first video data to train an animal information recognition model; the first labeled data including: basic information, behavioral characteristics, disease model, and biometrics; a test data acquisition module, used to acquire test video data and input it into the animal information recognition model for recognition to obtain test information recognition results, the test information recognition results including basic information, behavioral characteristics, disease model, and biometrics; and a test result analysis module, used to analyze the behavioral characteristics in the test information recognition results to filter action sequences with an accuracy rate lower than a preset threshold, and after correction, train the animal information recognition model to determine the trained animal information recognition model.
[0016] Thirdly, this application provides a network device, including: a memory, a transceiver, and a processor; wherein the memory is used to store a computer program; the transceiver is used to send and receive data under the control of the processor; and the processor is used to read the computer program in the memory and execute the method as described in the first aspect.
[0017] Fourthly, this application provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in the first aspect.
[0018] The beneficial effects of the technical solution provided in this application are:
[0019] The proposed solution can be applied to animal information recognition scenarios, enabling the acquisition of basic, behavioral, and disease information from animal videos. Specifically, this solution can acquire video data of an animal throughout its entire lifecycle and use corresponding basic information, behavioral characteristics, disease models, and biometrics as annotations for the video data. This annotation is then used to train an animal information recognition model. The trained model can then be used to directly acquire basic information, behavioral characteristics, disease models, and biometrics from the video, thereby enabling the acquisition of basic animal information and disease diagnosis. Specifically, this solution can acquire first video data of the animal's entire life cycle. This first video data corresponds to the animal's disease model. The video data includes a first video corresponding to the animal's spontaneous behavior and a second video corresponding to the animal's behavioral test. Based on the first video data, first labeled data is acquired to train the animal information recognition model. The first labeled data includes: basic information, behavioral characteristics, disease model, and biometrics. After training the animal information recognition model, test video data can be acquired and input into the animal information recognition model for recognition to obtain test information recognition results. The test information recognition results include basic information, behavioral characteristics, disease model, and biometrics. After obtaining the test information recognition results, correcting all the test information recognition results would be labor-intensive. Therefore, this solution can only analyze the behavioral characteristics in the test information recognition results and filter action sequences with an accuracy rate lower than a preset threshold. After correction, the animal information recognition model is trained based on the corrected test information to determine the trained animal information recognition model. After training the animal information recognition model, video data to be analyzed can be acquired and input into the trained model to determine the analysis results. These results include basic information, behavioral characteristics, disease models, and biometrics. This approach, based on video data, can not only analyze animal behavior but also acquire other information about the animal, facilitating subsequent processing. Furthermore, the information collected by this approach can be stored in a database for further analysis or to train other models. Attached Figure Description
[0020] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments of this application will be briefly introduced below.
[0021] Figure 1 is a flowchart illustrating an animal information processing method according to an embodiment of this application;
[0022] Figure 2 is a schematic diagram of the structure of an animal information processing device according to an embodiment of this application;
[0023] Figure 3 is a structural block diagram of a network device according to an embodiment of this application;
[0024] Figure 4 is a structural block diagram of a user equipment according to an embodiment of this application. Detailed Implementation
[0025] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals identify the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application.
[0026] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms, while “a plurality” refers to two or more, and other quantifiers are similarly understood. It should be further understood that the word “comprising” as used in this application’s specification means the presence of the stated feature, integer, step, operation, element, and / or component, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when we say an element is “connected” or “coupled” to another element, it can be directly connected or coupled to the other element, or there may be intermediate elements. Furthermore, “connected” or “coupled” as used herein can include wireless connection or wireless coupling. The word “and / or” as used herein describes the relationship between related objects, indicating that three relationships can exist; for example, A and / or B can represent: A alone, A and B simultaneously, and B alone. The character “ / ” generally indicates that the preceding and following related objects are in an “or” relationship.
[0027] The proposed solution can be applied to animal information recognition scenarios, enabling the acquisition of basic, behavioral, and disease information from animal videos. Specifically, this solution can acquire video data of an animal throughout its entire lifecycle and use corresponding basic information, behavioral characteristics, disease models, and biometrics as annotations for the video data. This annotation is then used to train an animal information recognition model. The trained model can then directly retrieve basic information, behavioral characteristics, disease models, and biometrics from the video, enabling functions such as obtaining basic animal information and disease diagnosis. Furthermore, the information collected by this solution can be stored in a database for analysis or for training other models.
[0028] The following is an overall introduction to the scheme of this application:
[0029] This solution aims to address the problems of cumbersome processes and incomplete information in current animal behavior analysis. With the development of posture estimation and deep learning, numerous animal behavior analysis systems have emerged and evolved in recent years, primarily extracting body features of interest from videos for behavioral analysis. For researchers in the biological field without a computer science background, this method is costly to learn, complex to process, and the information obtained is limited to behavioral data, with accuracy requiring improvement. This results in a mismatch between the time and manpower invested by researchers and the richness and accuracy of the information obtained. This solution, through visual understanding technology based on a full life-cycle database, can directly obtain basic animal information, behavioral information, disease information, and biological indicators from videos, significantly reducing the time researchers spend processing videos and obtaining more target information.
[0030] Currently, existing behavior analysis techniques mainly fall into two categories. The first involves extracting keypoints for behavior clustering analysis, using clustering models such as Keypoint-MoSeq, Behavior Atlas, and DeepLabCut. Researchers acquire videos, manually annotate keypoints, train models, perform supervised or unsupervised behavior clustering, and annotate behaviors to obtain behavior sequences. Databases based on this technology have been established, allowing researchers to directly parse videos to obtain behavior sequences. The second approach utilizes video understanding techniques for behavioral data analysis, employing deep neural networks such as DeepEthogram. Supervised machine learning networks are used to classify raw video streams into continuous behavior sequences.
[0031] This solution establishes a database based on visual understanding to reduce manual costs. Firstly, it employs visual understanding with verbal annotation, simplifying the process. Secondly, it collects behavioral videos of commonly used laboratory animal groups throughout their life cycles and trains them into a multimodal large-scale model. This model can automatically identify animal facial features, reducing the tedious process of manual annotation. As the database expands, the model can recognize various commonly used laboratory animals in most scenarios.
[0032] This approach enriches the full lifecycle database to improve the method's universality. By establishing a database of commonly used laboratory animals throughout their entire lifecycle, the method can be applied to most laboratory animal scenarios. Besides spontaneous behavior, it maintains high accuracy in commonly used task behaviors (such as behavioral testing) and scenarios involving carried devices.
[0033] This approach utilizes high-quality videos analyzed using existing technologies to train the model, thereby improving accuracy. First, existing technologies such as DeepLabcut and Behaviour Atlas are used for preliminary video analysis. Combined with human assistance, high-quality videos are selected and added to the database to ensure model accuracy. Then, the model's versatility is ensured by enriching the training content.
[0034] This approach utilizes algorithms and devices to assist in acquiring comprehensive information. In addition to video data, it integrates basic animal information, behavioral characteristics, disease models, and biomarkers into a database. With the aid of relevant algorithms and auxiliary devices, researchers can obtain animal information to the greatest extent possible.
[0035] The basic content of this technical solution is as follows: This solution uses a multimodal large language model to perform long-term behavioral video analysis on commonly used laboratory animals throughout their entire life cycle, generating textual descriptions of the animals' basic information, behavioral characteristics, disease models, and biomarkers. The technical solution includes five parts: data acquisition, model training, model tuning, data analysis, and system construction.
[0036] Data Acquisition: Data collection animals include commonly used animals such as mice, rats, dogs, and monkeys, as well as their disease models. Videos of at least 10 animals throughout their entire lifespan are collected for each disease model. Spontaneous behavior recordings last 1 hour, and behavioral test recordings last 5-15 minutes. Animals under 3 years old are recorded weekly; animals over 3 years old are recorded monthly. Basic information such as weight, heart rate, and respiratory rate are also collected. Video recording is divided into single-view and multi-view (multi-view is preferred). Multi-view recordings consist of four peripheral cameras and one top camera (single-view recordings only use the top camera). A checkerboard pattern is used for calibration during acquisition. After acquisition, DeepLabCut and Behaviour Atlas are used for preliminary analysis to exclude videos with inconsistent behavioral classifications.
[0037] Model Training: After filtering out cluttered videos, and referencing the preliminary results of behavioral analysis, text annotations were performed on the video content. The annotations included four parts: basic information, behavioral characteristics, disease model, and biomarkers. Basic information included animal species, sex, age, and weight. Behavioral characteristics included behavioral definition, duration, and behavioral parameters. Disease models included natural animals or disease model animals, disease course, and main manifestations. Biomarkers included heart rate, blood pressure, respiratory rate, and electroencephalogram (EEG). For example: a mouse, 8 weeks old, 22g; resting for 30s, sniffing for 10s, climbing for 3s, looking around for 4s, average speed 20mm / s; anesthetized mouse, 15 minutes post-anesthesia, movement speed less than 25mm / s; respiratory rate 85 breaths / min, heart rate 486 beats / min, oxygen consumption 1230 mm² / g live body weight, ventilation 14 ml / min, tidal volume 0.11 ml, systolic blood pressure 99 mmHg, diastolic blood pressure 66 mmHg.
[0038] Model tuning: After training the model using the dataset, new videos are collected as a test set to evaluate the model's accuracy. Action sequences with an accuracy below 80% are manually selected to form a short-time clip dataset, which is then subjected to filtering and manual correction.
[0039] Data collection and model training are carried out in the following order: first, spontaneous behavior of natural animals; then, task behavior of natural animals (including social behavior); then, spontaneous behavior of disease models and animal behavior; and finally, auxiliary devices (such as light recording, vital sign monitors, various invasive or non-invasive instruments, etc.) are added to the database step by step.
[0040] System Construction: Design a complete data analysis workflow, including basic data in four parts: basic information, behavioral information, disease models, and biomarkers, along with their correlations and spatial distribution characteristics. Construct a complete system to guide researchers in recording videos, parsing the videos, analyzing the data, and exporting the results.
[0041] Based on the above embodiments, this application also provides an animal information processing method, as shown in FIG1, the method comprising:
[0042] Step 102: Acquire the first video data of the animal's entire life cycle. The first video data corresponds to the animal's disease model. The video data includes a first video corresponding to the animal's spontaneous behavior and a second video corresponding to the animal's behavioral tests. As an optional embodiment, in the first video data of the animal's entire life cycle: the first video is recorded for 1 hour, and the second video is recorded for 5-15 minutes; the frequency of video collection for animals with a lifespan of less than 3 years is once a week; the frequency of video collection for animals over 3 years old is once a month; video recording is divided into single-view and multi-view; the multi-view consists of 4 peripheral cameras and 1 top camera; the single-view includes the top camera; basic information is collected when acquiring the first video data to form the first labeled data of the first video data based on the basic information. This solution can use long-term videos to analyze basic information, behavioral information, diseases, and biological characteristics, which can reduce the amount of labeled content.
[0043] Step 104: Based on the first video data, obtain the first labeled data to train the animal information recognition model; the first labeled data includes: basic information, behavioral characteristics, disease models, and biometrics. As an optional embodiment, the basic information includes animal species, sex, age, and weight; the behavioral characteristics include behavior definition, duration, and behavioral parameters; the disease model includes natural animals or disease model animals, disease course, and main manifestations; the biometrics include heart rate, blood pressure, respiratory rate, and electroencephalogram (EEG). As an optional embodiment, the first labeled data also includes sound information and ambient temperature information.
[0044] Step 106: Obtain test video data and input it into the animal information recognition model for recognition to obtain test information recognition results. The test information recognition results include basic information, behavioral characteristics, disease models, and biological indicators.
[0045] Step 108: Analyze the behavioral features in the test information recognition results to filter action sequences with an accuracy rate lower than a preset threshold, and after correction, train the animal information recognition model to determine the trained animal information recognition model.
[0046] The implementation methods of this application are similar to those of the above embodiments. For specific implementation methods, please refer to the specific implementation methods of the above embodiments, which will not be repeated here.
[0047] The proposed solution can be applied to animal information recognition scenarios, enabling the acquisition of basic, behavioral, and disease information from animal videos. Specifically, this solution can acquire video data of an animal throughout its entire lifecycle and use corresponding basic information, behavioral characteristics, disease models, and biometrics as annotations for the video data. This annotation is then used to train an animal information recognition model. The trained model can then be used to directly acquire basic information, behavioral characteristics, disease models, and biometrics from the video, thereby enabling the acquisition of basic animal information and disease diagnosis. Specifically, this solution can acquire first video data of the animal's entire life cycle. This first video data corresponds to the animal's disease model. The video data includes a first video corresponding to the animal's spontaneous behavior and a second video corresponding to the animal's behavioral test. Based on the first video data, first labeled data is acquired to train the animal information recognition model. The first labeled data includes: basic information, behavioral characteristics, disease model, and biometrics. After training the animal information recognition model, test video data can be acquired and input into the animal information recognition model for recognition to obtain test information recognition results. The test information recognition results include basic information, behavioral characteristics, disease model, and biometrics. After obtaining the test information recognition results, correcting all the test information recognition results would be labor-intensive. Therefore, this solution can only analyze the behavioral characteristics in the test information recognition results and filter action sequences with an accuracy rate lower than a preset threshold. After correction, the animal information recognition model is trained based on the corrected test information to determine the trained animal information recognition model.
[0048] Before annotating video data, this solution can input video data into a behavior analysis model to analyze behavior and remove low-quality videos with chaotic animal behavior. Furthermore, after training the animal information recognition model, this solution can acquire the video data to be analyzed and input it into the trained animal information recognition model to determine the animal's analysis results. These analysis results include basic information, behavioral characteristics, disease models, and biometrics. This solution can not only analyze animal behavior based on video data but also obtain other information about the animal, facilitating subsequent processing. Specifically, as an optional embodiment, before acquiring the video annotation data, the method further includes: inputting the video data into the behavior analysis model to determine the behavior analysis results, and removing low-quality videos from the video data based on the behavior analysis results; the method also includes: acquiring the video data to be analyzed and inputting it into the trained animal information recognition model to determine the animal's analysis results, which include basic information, behavioral characteristics, disease models, and biometrics. The model training process can include: inputting data into the model, the model obtaining analysis results, and adjusting the model parameters based on the differences between the analysis results and the annotations, thereby obtaining a trained model.
[0049] In training the animal information recognition model, this solution can initially train the model based on existing videos and annotations of disease-free animals, and then train the model based on videos and annotations of diseased animals. Specifically, as an optional embodiment, the steps for training the animal information recognition model include: acquiring second video data of animals and corresponding annotations, the second video data including videos of disease-free animals, a third video corresponding to spontaneous animal behavior, and a fourth video corresponding to animal behavior tests; acquiring the third video and its corresponding annotations, and training the animal information recognition model; acquiring the fourth video and its corresponding annotations, and training the animal information recognition model trained on the third video; acquiring the first video and its corresponding annotations, and training the animal information recognition model trained on the fourth video; acquiring the second video and its corresponding annotations, and training the animal information recognition model trained on the first video. Taking data sequentially for model training in a set order can improve the model's recognition accuracy.
[0050] Existing video data has various annotation sources: some are manually annotated, some are obtained through video analysis, and some are obtained from acquisition devices. Therefore, this solution can first train the model based on low-quality annotations (such as those obtained through video analysis, which may be inaccurate) and videos, and then fine-tune the model using manual annotations or annotations obtained from acquisition devices. Specifically, as an optional embodiment, the steps for training the animal information recognition model include: obtaining a fifth video and its corresponding annotations, the annotations for the fifth video being obtained through analysis of the fifth video; obtaining a sixth video and its corresponding annotations, the annotations for the sixth video being obtained through data collection of animals using an auxiliary acquisition device; training the animal information recognition model based on the fifth video and its annotations, and training the animal information recognition model trained on the fifth video based on the sixth video and its annotations.
[0051] After obtaining the test information recognition results, correcting all the test information recognition results would be labor-intensive. Therefore, this solution can only analyze the behavioral features in the test information recognition results. Specifically, as an optional embodiment, the analysis of the behavioral features in the test information recognition results to filter action sequences with an accuracy rate lower than a preset threshold, and after correction, training the animal information recognition model to determine the trained animal information recognition model, includes: analyzing the behavioral features in the test information recognition results by manual screening to filter action sequences with an accuracy rate lower than a preset threshold; filtering and correcting the selected action sequences and performing manual correction to determine test update data; and adjusting the animal information recognition model based on the test update data to determine the trained animal information recognition model.
[0052] Based on the above embodiments, this application also provides an animal information processing device, as shown in FIG2, the device comprising:
[0053] The video data acquisition module 202 is used to acquire first video data of the animal throughout its entire life cycle. The first video data corresponds to the animal's disease model. The video data includes a first video corresponding to the animal's spontaneous behavior and a second video corresponding to the animal's behavior test.
[0054] The recognition model training module 204 is used to obtain first labeled data based on the first video data in order to train the animal information recognition model; the first labeled data includes: basic information, behavioral characteristics, disease model, and biological indicators.
[0055] The test data acquisition module 206 is used to acquire test video data and input it into the animal information recognition model for recognition, so as to obtain test information recognition results. The test information recognition results include basic information, behavioral characteristics, disease models, and biological indicators.
[0056] The test result analysis module 208 is used to analyze the behavioral features in the test information recognition results, to filter action sequences with an accuracy rate lower than a preset threshold, and after correction, to train the animal information recognition model to determine the trained animal information recognition model.
[0057] The implementation methods of this application are similar to those of the above embodiments. For specific implementation methods, please refer to the specific implementation methods of the above embodiments, which will not be repeated here.
[0058] The proposed solution can be applied to animal information recognition scenarios, enabling the acquisition of basic, behavioral, and disease information from animal videos. Specifically, this solution can acquire video data of an animal throughout its entire lifecycle and use corresponding basic information, behavioral characteristics, disease models, and biometrics as annotations for the video data. This annotation is then used to train an animal information recognition model. The trained model can then be used to directly acquire basic information, behavioral characteristics, disease models, and biometrics from the video, thereby enabling the acquisition of basic animal information and disease diagnosis. Specifically, this solution can acquire first video data of the animal's entire life cycle. This first video data corresponds to the animal's disease model. The video data includes a first video corresponding to the animal's spontaneous behavior and a second video corresponding to the animal's behavioral test. Based on the first video data, first labeled data is acquired to train the animal information recognition model. The first labeled data includes: basic information, behavioral characteristics, disease model, and biometrics. After training the animal information recognition model, test video data can be acquired and input into the animal information recognition model for recognition to obtain test information recognition results. The test information recognition results include basic information, behavioral characteristics, disease model, and biometrics. After obtaining the test information recognition results, correcting all the test information recognition results would be labor-intensive. Therefore, this solution can only analyze the behavioral characteristics in the test information recognition results and filter action sequences with an accuracy rate lower than a preset threshold. After correction, the animal information recognition model is trained based on the corrected test information to determine the trained animal information recognition model. After training the animal information recognition model, video data to be analyzed can be acquired and input into the trained model to determine the analysis results. These results include basic information, behavioral characteristics, disease models, and biometrics. This solution, based on video data, can not only analyze animal behavior but also obtain other information about the animal, facilitating subsequent processing.
[0059] It should be noted that the division of units and / or modules in the embodiments of this application is illustrative and only represents one logical functional division. In actual implementation, there may be other division methods. Furthermore, the functional units and / or modules in the various embodiments of this application can be integrated into one processing unit and / or module, or each unit and / or module can exist physically separately, or two or more units and / or modules can be integrated into one unit and / or module. The integrated units and / or modules described above can be implemented in hardware or as software functional units and / or modules.
[0060] If the integrated units and / or modules are implemented as software functional units and / or modules and sold or used as independent products, they can be stored in a processor-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to related technologies, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0061] Furthermore, the data transmission apparatus and data transmission method provided in the above embodiments are based on the same application concept. Since the methods and apparatus solve problems in similar principles, the implementation of the apparatus and methods can refer to each other, and repeated parts will not be described again.
[0062] Figure 3 is a structural block diagram of a network device according to an exemplary embodiment.
[0063] As shown in Figure 3, the network device 1100 includes at least: a processor 1110, a memory 1120, and a transceiver 1130.
[0064] The transceiver 1130 is used to receive and send data under the control of the processor 1110.
[0065] In Figure 3, the bus architecture may include any number of interconnected buses and bridges, specifically linking various circuits of one or more processors represented by processor 1110 and memory represented by memory 1120. The bus architecture may also link various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. The bus interface provides an interface. Transceiver 1130 may be multiple elements, including transmitters and receivers, providing units and / or modules for communicating with various other devices over transmission media, including wireless channels, wired channels, optical fibers, and other transmission media.
[0066] The processor 1110 is responsible for managing the bus architecture and general processing, and the memory 1120 can store the data used by the processor 1110 when performing operations.
[0067] Optionally, the processor 1110 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or a complex programmable logic device (CPLD). The processor 1110 may also adopt a multi-core architecture. The processor 1110 and the memory 1120 may also be physically separated.
[0068] The processor 1110 calls the computer program stored in the memory 1120 to execute any of the cell wireless network temporary identifier allocation methods provided in the above embodiments of this application according to the obtained executable instructions.
[0069] Figure 4 is a structural block diagram of a user equipment according to an exemplary embodiment.
[0070] As shown in Figure 4, the user equipment 1300 includes at least: a processor 1310, a memory 1320, and a transceiver 1330.
[0071] The transceiver 1330 is used to receive and send data under the control of the processor 1310.
[0072] In Figure 4, the bus architecture may include any number of interconnected buses and bridges, specifically linking various circuits of one or more processors represented by processor 1310 and memory represented by memory 1320. The bus architecture may also link various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. The bus interface provides an interface. Transceiver 1330 may be multiple elements, including transmitters and receivers, providing units and / or modules for communicating with various other devices over transmission media, including wireless channels, wired channels, optical fibers, etc. For different user equipment, user interface 1340 may also be an interface capable of connecting external or internal devices, including but not limited to keypads, displays, speakers, microphones, joysticks, etc.
[0073] The processor 1310 is responsible for managing the bus architecture and general processing, and the memory 1320 can store the data used by the processor 1310 when performing operations.
[0074] Optionally, the processor 1310 can be a CPU (Central Processing Unit), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or a CPLD (Complex Programmable Logic Device). The processor 1310 can also adopt a multi-core architecture. The processor 1310 and the memory 1320 can also be physically separated.
[0075] The processor 1310 calls the computer program stored in the memory 1320 to execute any of the cell wireless network temporary identifier allocation methods provided in the above embodiments of this application according to the obtained executable instructions.
[0076] It should be noted that the apparatus provided in this application embodiment can implement all the method steps implemented in the above method embodiment and can achieve the same technical effect. Here, the parts that are the same as those in the method embodiment and the beneficial effects will not be described in detail.
[0077] Furthermore, this application provides a storage medium storing a computer program, which, when executed by a processor, implements the data transmission methods described in the above embodiments. The storage medium can be any available medium or data storage device accessible to the processor, including but not limited to magnetic storage (e.g., floppy disks, hard disks, magnetic tapes, magneto-optical disks (MO), etc.), optical storage (e.g., CDs, DVDs, BDs, HVDs, etc.), and semiconductor storage (e.g., ROMs, EPROMs, EEPROMs, non-volatile memory (NAND flash), solid-state drives (SSDs)).
[0078] This application provides a program product, such as an FPGA chip or a DSP chip, which includes executable instructions stored in a storage medium. A processor reads the executable instructions from the storage medium, causing the processor to execute the executable instructions to implement the data transmission methods described in the above embodiments.
[0079] The proposed solution can be applied to animal information recognition scenarios, enabling the acquisition of basic, behavioral, and disease information from animal videos. Specifically, this solution can acquire video data of an animal throughout its entire lifecycle and use corresponding basic information, behavioral characteristics, disease models, and biometrics as annotations for the video data. This annotation is then used to train an animal information recognition model. The trained model can then be used to directly acquire basic information, behavioral characteristics, disease models, and biometrics from the video, thereby enabling the acquisition of basic animal information and disease diagnosis. Specifically, this solution can acquire first video data of the animal's entire life cycle. This first video data corresponds to the animal's disease model. The video data includes a first video corresponding to the animal's spontaneous behavior and a second video corresponding to the animal's behavioral test. Based on the first video data, first labeled data is acquired to train the animal information recognition model. The first labeled data includes: basic information, behavioral characteristics, disease model, and biometrics. After training the animal information recognition model, test video data can be acquired and input into the animal information recognition model for recognition to obtain test information recognition results. The test information recognition results include basic information, behavioral characteristics, disease model, and biometrics. After obtaining the test information recognition results, correcting all the test information recognition results would be labor-intensive. Therefore, this solution can only analyze the behavioral characteristics in the test information recognition results and filter action sequences with an accuracy rate lower than a preset threshold. After correction, the animal information recognition model is trained based on the corrected test information to determine the trained animal information recognition model. After training the animal information recognition model, video data to be analyzed can be acquired and input into the trained model to determine the analysis results. These results include basic information, behavioral characteristics, disease models, and biometrics. This solution, based on video data, can not only analyze animal behavior but also obtain other information about the animal, facilitating subsequent processing.
[0080] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.
[0081] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer-executable instructions. These computer-executable instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in one or more blocks of the flowchart illustrations and / or one or more blocks of the block diagrams.
[0082] These processor-executable instructions may also be stored in a processor-readable memory that can instruct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the processor-readable memory produce an article of manufacture including instruction means that implement the functions specified in one or more flowcharts and / or one or more block diagrams.
[0083] These processor-executable instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, such that the instructions, which execute on the computer or other programmable apparatus, provide steps for implementing the functions specified in one or more flowcharts and / or one or more block diagrams.
[0084] It should be understood that although the steps in the flowcharts of the accompanying figures are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the accompanying figures may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.
[0085] The above description is only a partial embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.
Claims
1. An animal information processing method, characterized in that, The method includes: Acquire first video data of the animal throughout its entire life cycle. The first video data corresponds to the animal's disease model. The video data includes a first video corresponding to the animal's spontaneous behavior and a second video corresponding to the animal's behavioral test. Based on the first video data, the first labeled data is obtained to train the animal information recognition model; the first labeled data includes: basic information, behavioral characteristics, disease models, and biological indicators. The test video data is acquired and input into the animal information recognition model for recognition, and the test information recognition results are obtained. The test information recognition results include basic information, behavioral characteristics, disease models, and biological indicators. The behavioral features in the test information recognition results are analyzed to filter action sequences with an accuracy rate lower than a preset threshold. After correction, the animal information recognition model is trained to determine the well-trained animal information recognition model.
2. The method as described in claim 1, characterized in that, Before acquiring video annotation data, the method further includes: The video data is input into the behavior analysis model to determine the behavior analysis results, and low-quality videos are removed from the video data based on the behavior analysis results. The method further includes: The video data to be analyzed is acquired and input into a trained animal information recognition model to determine the analysis results of the animal. The analysis results include basic information, behavioral characteristics, disease models, and biometrics.
3. The method as described in claim 1, characterized in that, The steps for training an animal information recognition model include: Acquire second video data of animals and corresponding annotations. The second video data includes videos of disease-free animals, and also includes a third video corresponding to spontaneous animal behavior and a fourth video corresponding to animal behavior tests. Obtain the third video and its corresponding annotations, and train the animal information recognition model. Obtain the fourth video and its corresponding annotations, and train the animal information recognition model trained on the third video. Obtain the first video and its corresponding annotations, and train the animal information recognition model trained on the fourth video. Obtain the second video and its corresponding annotations, and train the animal information recognition model trained on the first video.
4. The method as described in claim 1, characterized in that, The steps for training an animal information recognition model include: The fifth video and its corresponding annotation were obtained by analyzing the fifth video. The sixth video and its corresponding annotation were obtained. The annotation for the sixth video was obtained after data collection of the animal using an auxiliary data acquisition device. The animal information recognition model was trained based on the fifth video and its annotations, and the animal information recognition model trained on the fifth video was trained based on the sixth video and its annotations.
5. The method as described in claim 1, characterized in that, The basic information includes animal species, sex, age, and weight; the behavioral characteristics include behavioral definition, duration, and behavioral parameters; the disease model includes natural animals or disease model animals, disease course, and main manifestations; and the bioindicators include heart rate, blood pressure, respiratory rate, and electroencephalogram (EEG).
6. The method as described in claim 1, characterized in that, The first video data of the animal's entire life cycle: the first video recording time is 1 hour, and the second video recording time is 5-15 minutes; the video collection frequency for animals with a lifespan of less than 3 years is once a week; the video collection frequency for animals with a lifespan of more than 3 years is once a month; the video recording is divided into single-view and multi-view; the multi-view consists of 4 surrounding cameras and 1 top camera; The single-view camera includes a top camera; when acquiring the first video data, basic information is collected to form the first annotation data of the first video data based on the basic information.
7. The method as described in claim 1, characterized in that, The first labeled data also includes sound information and ambient temperature information.
8. The method as described in claim 1, characterized in that, The process of analyzing behavioral features in the test information recognition results to filter action sequences with accuracy below a preset threshold, and then training the animal information recognition model after correction to determine the trained animal information recognition model, includes: Based on manual screening, the behavioral characteristics in the test information recognition results are analyzed to filter action sequences with an accuracy rate lower than a preset threshold. The selected action sequences are filtered and manually corrected to determine the test update data. Based on this updated data, the animal information recognition model is adjusted to determine the best-trained animal information recognition model. Information recognition model.
9. An animal information processing device, characterized in that, The device includes: The video data acquisition module is used to acquire the first video data of the animal throughout its entire life cycle. The first video data corresponds to the animal's disease model. The video data includes the first video corresponding to the animal's spontaneous behavior and the second video corresponding to the animal's behavior test. The recognition model training module is used to obtain first labeled data based on the first video data in order to train the animal information recognition model; the first labeled data includes: basic information, behavioral characteristics, disease models, and biological indicators. The test data acquisition module is used to acquire test video data and input it into the animal information recognition model for recognition, so as to obtain test information recognition results, which include basic information, behavioral characteristics, disease models, and biological indicators. The test result analysis module is used to analyze the behavioral features in the test information recognition results, to filter action sequences with an accuracy rate lower than a preset threshold, and after correction, to train the animal information recognition model to determine the trained animal information recognition model.
10. A network device, characterized in that, include: The system includes a memory, a transceiver, and a processor; wherein the memory is used to store computer programs; and the transceiver is used to send and receive data under the control of the processor. The processor is configured to read the computer program in the memory and execute the method as described in claims 1-8.