Cashmere goat identifying and tracking method and system based on large model

Abstract

The invention provides a method and a system for identifying and tracking down producing goats based on a large model. The method comprises the following steps: determining group motion state information of down producing goats in a polyculture area based on monitoring video data of the down producing goat polyculture area; if the group motion state information meets a preset condition, intercepting a first group of continuous video frames containing face information of each cashmere goat from the monitoring video data, and performing identity recognition by using a first model to obtain a first conclusion whether the identity information of the corresponding cashmere goat belongs to a double-lamb breeding identity; otherwise, intercepting a second group of continuous video frames containing the face information of each cashmere goat from the monitoring video data, and performing identity recognition by using a second model to obtain a second conclusion whether the identity information of the corresponding cashmere goat belongs to the double-lamb breeding identity or not. According to the method for identifying and tracking the down producing goats based on the combination of the large model and the non-large model, the identity identification requirements of the down producing goats in different activity states can be flexibly met.

Description

Technical Field

[0001] The present invention relates to the technical field of breeding of cashmere goats, and in particular, to a method and system for identifying and tracking cashmere goats based on a large model. Background Art

[0002] The breeding of twin lambs in cashmere goats is an important breeding work aimed at improving the lambing rate of cashmere goats and obtaining more economic benefits. Among them, for basic ewes, healthy adult ewes that have given birth to twins or even multiple lambs in the past are preferentially selected from flocks with complete production records. They are required to have a moderate body size, strong physique, fine and dense cashmere, stable production, well-developed reproductive organs, and no history of reproductive diseases; for rams, in addition to being physically strong and having strong libido, it is necessary to trace the reproductive performance of their maternal families, and rams with more lambs born to their maternal lines are preferentially selected.

[0003] Before the selected rams and ewes are grouped and separately raised, it is necessary to identify and track the selected rams and ewes so that when group separation and individual raising are required, it is possible to accurately identify which sheep are the selected rams and ewes, which requires the identification of sheep identities. Currently, using image recognition technology to identify target sheep is a relatively popular technology, mainly based on sheep face recognition to determine the identity of sheep. However, in the prior art, when performing sheep face recognition, it is mainly based on static sheep face images, but the faces of sheep are often in a moving state (such as eating, ruminating) and running state. Simply identifying the identity of sheep based on static sheep face images has obvious deficiencies in accuracy.

[0004] Therefore, how to more accurately identify the identity of sheep to quickly identify the rams and ewes used for twin lamb breeding is a technical problem that needs to be solved currently. Summary of the Invention

[0005] In order to solve the technical problems existing in the above background art, the present invention provides a method, system, electronic device, computer storage medium, and computer program product for identifying and tracking cashmere goats based on a large model.

[0006] The present invention provides a method for identifying and tracking cashmere goats based on a large model, and the method includes the following steps: Determine the group motion state information of cashmere goats in the mixed breeding area based on the monitoring video data of the cashmere goat mixed breeding area, and judge whether the group motion state information meets a preset condition; If the group motion state information meets the preset conditions, the first model is called, the first model is constructed based on the large model, and is obtained based on the facial image training of the cashmere goats in motion; a first group of continuous video frames containing facial information of each cashmere goat is intercepted from the monitoring video data, and the first model is used to perform identity recognition on the first group of continuous video frames to obtain a first conclusion on whether the identity information of the corresponding cashmere goat belongs to the twin lamb breeding identity; If the group motion status information does not meet the preset conditions, the second model is called, which is constructed based on a non-large model and is trained based on static facial images and dynamic facial images of cashmere goats in static state; a second group of continuous video frames containing facial information of each cashmere goat is intercepted from the monitoring video data, and the second model is used to perform identity recognition on the second group of continuous video frames to obtain a second conclusion on whether the identity information of the corresponding cashmere goat belongs to the identity of twin lamb breeding.

[0007] Optionally, determining group motion state information of cashmere goats in the mixed-breeding area based on surveillance video data of the cashmere goats in the mixed-breeding area, and judging whether the group motion state information meets a preset condition includes: Determine the group movement state information of cashmere goats in the mixed-breeding area based on the surveillance video data of the cashmere goat mixed-breeding area; When the group motion state information corresponds to a group motion state, further determining the cause factor causing the group motion state based on the monitoring video data; if the cause factor can be determined, determining that the preset condition is not satisfied; if the cause factor cannot be determined, determining that the preset condition is satisfied; When the group motion state information corresponds to a non-group motion state, it is determined that the preset condition is not satisfied.

[0008] Optionally, the first model is trained in the following manner: Acquire facial images of cashmere goats in motion under different lighting conditions and for different reasons; wherein the reasons at least include running and fighting; Acquire static facial images and dynamic facial images of cashmere goats under static and corresponding lighting conditions, and construct the facial images, static facial images, and dynamic facial images belonging to the same cashmere goat under the same lighting conditions into one piece of training data, thereby obtaining a training data set; The large model is fine-tuned using the training data set to obtain the first model.

[0009] Optionally, when the cause factor can be determined and judged not to meet the preset condition, it also includes: Calculating the number of the cause factors and determining the waiting time according to the number; wherein, when the number is greater, the waiting time is set longer, and vice versa, the waiting time is set shorter; After reaching the waiting duration, second consecutive video frames containing face information of each cashmere goat are intercepted from the monitored video data, and the second model is used to perform identity recognition on the second consecutive video frames to obtain a second conclusion on whether the identity information of the corresponding cashmere goat belongs to the identity of twin lamb breeding.

[0010] Optionally, after reaching the waiting duration, intercepting second consecutive video frames containing face information of each cashmere goat from the monitored video data includes: Based on the monitored video data of the cashmere goat mixed breeding area, determining the latest group movement state information of the cashmere goats in the mixed breeding area, and determining whether the latest group movement state information meets a preset condition; If not, intercept the second consecutive video frames containing face information of each cashmere goat from the monitored video data.

[0011] The present invention also provides a cashmere goat identification and tracking system based on a large model. The system includes a processor and a memory, and the memory stores executable program code; the processor calls the executable program code stored in the memory to implement the following steps: Based on the monitored video data of the cashmere goat mixed breeding area, determining the group movement state information of the cashmere goats in the mixed breeding area, and determining whether the group movement state information meets a preset condition; If the group movement state information meets the preset condition, call the first model. The first model is constructed based on a large model and trained based on the face images of cashmere goats in a dynamic state; intercept the first consecutive video frames containing face information of each cashmere goat from the monitored video data, and use the first model to perform identity recognition on the first consecutive video frames to obtain a first conclusion on whether the identity information of the corresponding cashmere goat belongs to the identity of twin lamb breeding; If the group movement state information does not meet the preset condition, call the second model. The second model is constructed based on a non-large model and trained based on the static face images and dynamic face images of cashmere goats in a static state; intercept the second consecutive video frames containing face information of each cashmere goat from the monitored video data, and use the second model to perform identity recognition on the second consecutive video frames to obtain a second conclusion on whether the identity information of the corresponding cashmere goat belongs to the identity of twin lamb breeding.

[0012] Optionally, based on the monitored video data of the cashmere goat mixed breeding area, determining the group movement state information of the cashmere goats in the mixed breeding area, and determining whether the group movement state information meets a preset condition includes: Based on the monitored video data of the cashmere goat mixed breeding area, determining the group movement state information of the cashmere goats in the mixed breeding area; When the group motion state information corresponds to the group motion state, further determine the cause factors leading to the group motion state based on the monitoring video data; if the cause factors can be determined, it is determined that the preset conditions are not met; if the cause factors cannot be determined, it is determined that the preset conditions are met; When the group motion state information corresponds to the non-group motion state, it is determined that the preset conditions are not met.

[0013] Optionally, the first model is trained in the following manner: Obtain the face images of cashmere goats in different motion states under different lighting conditions and for different reasons; where the reasons include at least running and fighting; Obtain the face static images and face dynamic images of cashmere goats under static and corresponding lighting conditions, and construct the face images, face static images, and face dynamic images of the same cashmere goat under the same lighting conditions into a piece of training data, so as to obtain a training data set; Use the training data set to fine-tune and train the large model to obtain the first model.

[0014] Optionally, when the cause factors can be determined and it is determined that the preset conditions are not met, it further includes: Calculate the number of the cause factors, and determine the waiting duration according to the number; where when the number is larger, the waiting duration is set longer, and vice versa, the waiting duration is set shorter; After reaching the waiting duration, intercept the second set of consecutive video frames containing the face information of each cashmere goat from the monitoring video data, and use the second model to perform identity recognition on the second set of consecutive video frames to obtain a second conclusion on whether the identity information of the corresponding cashmere goat belongs to the identity of dual-lamb breeding.

[0015] Optionally, after reaching the waiting duration, intercept the second set of consecutive video frames containing the face information of each cashmere goat from the monitoring video data, including: Based on the monitoring video data of the cashmere goat mixed breeding area again, determine the latest group motion state information of the cashmere goats in the mixed breeding area, and judge whether the latest group motion state information meets the preset conditions; If not, intercept the second set of consecutive video frames containing the face information of each cashmere goat from the monitoring video data.

[0016] The present invention also provides an electronic device, including: a memory storing executable program code; a processor coupled to the memory; the processor calls the executable program code stored in the memory and executes the method described in any one of the above.

[0017] The present invention also provides a computer storage medium, on which a computer program is stored, and when the computer program is run by a processor, it executes the method described in any one of the above.

[0018] The present invention also provides a computer program product, which includes a computer program stored in a computer storage medium, and when the computer program is executed by a processor of an electronic device, it implements the method described in any one of the above.

[0019] The method for identifying and tracking cashmere goats based on the combination of large models and non-large models of the present invention can flexibly meet the identity recognition requirements of cashmere goats in different activity states. When the flock is active, it accurately tracks with the powerful dynamic feature learning ability of the large model; when the flock is stationary, it efficiently and stably identifies with a lightweight non-large model. It not only improves the accuracy of identifying target sheep in the double-lamb breeding process, reduces misjudgment, but also takes into account the reasonable allocation of computing resources in different scenarios, reduces the overall recognition cost and time consumption, and ensures the efficient and orderly development of the breeding work. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments recorded in the present application. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0021] Figure 1 It is a schematic diagram of a method for identifying and tracking cashmere goats based on a large model disclosed in an embodiment of the present invention.

[0022] Figure 2 It is a schematic diagram of a set of static facial images of cashmere goats in a static state disclosed in an embodiment of the present invention.

[0023] Figure 3 It is a schematic diagram of a system for identifying and tracking cashmere goats based on a large model disclosed in an embodiment of the present invention.

[0024] It should be noted that although the steps in the flowchart diagrams (if any) involved in each embodiment are drawn in sequence according to the indication of the arrows, unless there is a clear description in this article, the execution of these steps does not have a strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowchart diagrams involved in each embodiment may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same moment, but can be executed at different moments, and the execution order of these steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or steps or stages in other steps. Detailed implementation manners

[0025] In order to make the objectives, technical solutions, and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be construed as limiting the present application. All other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present application.

[0026] Terms such as "first", "second", "third", "fourth", etc. (if any) in the specification, claims, and drawings of the present application are used to distinguish similar objects and do not necessarily describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily limit to those clearly listed steps or units, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0027] In the following description, references to "a specific implementation manner" or "a specific example" and the like describe subsets of all possible embodiments. However, it can be understood that "a specific implementation manner" or "a specific example" can be the same subset or different subsets of all possible embodiments and can be combined with each other without conflict. In the following description, the term "a plurality" refers to at least two. When it is said in the present application that a certain value reaches a threshold (if any), in some specific examples, it may include the case where the former is greater than the latter.

[0028] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0029] Please refer to Figure 1 , an embodiment of the present invention discloses a method for identifying and tracking cashmere goats based on a large model. The method includes the following steps: Determine the group motion state information of the cashmere goats in the mixed breeding area based on the monitoring video data of the cashmere goat mixed breeding area, and determine whether the group motion state information meets a preset condition.

[0030] In this step, multiple monitoring cameras are installed in the mixed grazing area of cashmere goats. From the monitoring video data collected by these cameras, the information on the group movement state of cashmere goats in the mixed grazing area can be obtained, including the overall moving speed, moving direction, aggregation or dispersion degree of the flock, etc.

[0031] For example, by analyzing the video footage, if it is found that the flock quickly moves from one corner of the mixed grazing area to another corner within a short period (such as within 1 minute), and the moving speed exceeds the normal walking speed, reaching about 1 meter per second, this is an obvious group movement state; or, the flock that was originally tightly gathered together for feeding suddenly scatters and runs in all directions. These situations all constitute the information on the group movement state. Using image recognition algorithms and video analysis techniques, the video frames are parsed frame by frame to extract the dynamic changes in the contours of each sheep, and then the above-mentioned group movement state parameters are statistically analyzed to determine the information on the group movement state of cashmere goats in the current mixed grazing area. If it meets the above two situations, it is determined that the information on the group movement state meets the preset conditions; otherwise, it is determined that the information on the group movement state does not meet the preset conditions.

[0032] If the information on the group movement state meets the preset conditions, the first model is called. The first model is constructed based on large models (such as GPT, BERT) and trained based on the facial images of cashmere goats in motion; the first set of consecutive video frames containing facial information of each cashmere goat is intercepted from the monitoring video data, and the first model is used to perform identity recognition on this first set of consecutive video frames to obtain the first conclusion on whether the identity information of the corresponding cashmere goat belongs to the identity of twin lamb breeding.

[0033] In this step, the preset conditions can be set as the active state where the flock is in rapid movement and frequent position changes, such as the two situations mentioned above. The present invention constructs the first model based on a large model. The first model is trained based on the facial images of cashmere goats during running, and it can perform sheep face recognition on cashmere goats based on the special facial features of cashmere goats during running (different from the facial features in the static state), so as to obtain the first conclusion on whether the identity information of the corresponding cashmere goat belongs to the identity of twin lamb breeding.

[0034] The above-mentioned large model is, for example, a large-scale model based on the Transformer architecture, and is fine-tuned and trained using lightweight facial image data of cashmere goats in motion. These training facial image data come from cashmere goats under different lighting conditions, angles, and motion postures.

[0035] From the active monitoring video data, intercept the first set of consecutive video frames of each cashmere goat, for example, intercept video frames with a duration of 3 - 5 seconds and 15 frames per second. Input these video frames containing rich dynamic facial expressions and posture changes into the first model. The first model will focus on key facial feature points, such as the dynamic changes in the eye contours and the shape of the mouth and nose, compare them with the learned feature library, and determine whether the identity information of this goat matches the identity registered in the twin lamb breeding plan, and obtain the first conclusion.

[0036] If the group motion state information does not meet the preset conditions, then call the second model. The second model is constructed based on non - large models and is trained based on the static facial images and dynamic facial images of cashmere goats in a static state; intercept the second set of consecutive video frames containing facial information of each cashmere goat from the monitoring video data, and use the second model to perform identity recognition on this second set of consecutive video frames to obtain the second conclusion on whether the identity information of the corresponding cashmere goat belongs to the twin lamb breeding identity.

[0037] In this step, when the cashmere goats in the monitoring screen are in a relatively static state, such as during the collective noon rest period in the sheep pen, most of the goats stand or lie still, with only occasional head swings. At this time, the group motion state is relatively stable, that is, it does not meet the preset conditions. At this time, use non - large models such as traditional machine learning models (such as support vector machines, CNNs, etc.) to construct. The training data incorporates standard frontal and side static facial images of cashmere goats in a static state, as well as a small amount of facial dynamic images in slow motion (such as eating, ruminating, yawning, etc.).

[0038] Intercept the second set of consecutive video frames of each cashmere goat from the monitoring video data in a static state. This set of video frames has a shorter duration, such as 1 - 2 seconds, and the frame rate can also be reduced to 5 frames per second. Input these consecutive video frames into the second model. The second model relies on features such as the texture of static and dynamic faces and fixed facial feature ratios to judge the identity of the cashmere goat and output the second conclusion.

[0039] The cashmere goat recognition and tracking method based on the combination of large models and non - large models of the present invention can flexibly meet the identity recognition requirements of cashmere goats in different activity states. When the flock is active, it accurately tracks with the powerful dynamic feature learning ability of the large model; when the flock is static, it efficiently and stably identifies the identity of cashmere goats relying on lightweight non - large models. It not only improves the accuracy of identifying the identity of target sheep in the twin lamb breeding process, reduces misjudgment, but also takes into account the reasonable allocation of computing resources in different scenarios, reduces the overall recognition cost and time consumption, and ensures the efficient and orderly development of the breeding work.

[0040] Optionally, determining the group motion state information of cashmere goats in the mixed - raising area based on the monitoring video data of the cashmere goat mixed - raising area, and judging whether the group motion state information meets the preset conditions includes: Determine the group movement state information of the cashmere goats in the mixed breeding area based on the monitoring video data of the cashmere goat mixed breeding area; When the group movement state information corresponds to the group movement state, further determine the cause factors leading to the group movement state based on the monitoring video data; if the cause factors can be determined, it is determined that the preset conditions are not met; if the cause factors cannot be determined, it is determined that the preset conditions are met; When the group movement state information corresponds to a non-group movement state, it is determined that the preset conditions are not met.

[0041] In this embodiment, the monitoring video is like an "eye" that continuously records the activities of cashmere goats, and the video frames are analyzed frame by frame through computer vision technology. For example, the contour of each cashmere goat is identified using a target detection algorithm, and then the position changes of the contour in consecutive frames are tracked to calculate the overall movement speed, direction of the flock, and the change in the distance between goats. These parameters are summarized to form the group movement state information. If it is found that the flock is moving steadily in one direction or frequently shuttling disorderly in a certain area, it is determined that the flock is in a group movement state.

[0042] When it is monitored that the cashmere goats are not in a group movement state, for example, most of the goats are lying down and resting in the sheepfold, and only a few goats occasionally shake their heads and wag their tails, it is directly determined that the preset conditions are not met. This is because the facial postures of the goats in this state are relatively single and lack sufficient dynamic changes, making it more suitable to use the second model to identify the identity information of each cashmere goat.

[0043] When it is monitored that the cashmere goats are in a group movement state, that is, the preset conditions are met, the present invention sets to determine the cause factors leading to the group movement state from the monitoring video data, that is, to judge whether it can be determined which specific reasons cause the group movement of the flock. These cause factors are mainly related to the feeder and the administrator. The feeder drives the feed truck into the area, and the flock gathers and follows, or the administrator suddenly enters the mixed breeding area, causing the flock to be frightened. At this time, it can be determined that the feeder or the feed truck, the administrator is the determinable cause factor. If the above cause factors can be determined, it indicates that the flock will end the group movement state in a short time (because the feeder and the administrator will leave soon, or the flock will quickly adapt to the presence of the feeder and the administrator), but instead turn into a static state. At this time, it is determined that the preset conditions are not met, and the second model is called to perform sheep face recognition after a short time; if the above cause factors cannot be determined, it means that the movement of the flock is spontaneous (such as sensing abnormal sounds, there are restless goats, etc.), then it cannot be determined whether the flock will end the group movement state in a short time. At this time, it is determined that the preset conditions are met, and the first model is called to start sheep face recognition immediately.

[0044] Therefore, by determining whether the above-mentioned cause factors can be obtained, the present invention dynamically determines a more appropriate sheep face recognition model, thereby reducing the computing load of sheep face recognition.

[0045] Optionally, the first model is trained in the following manner: Acquire facial images of cashmere goats in motion under different lighting conditions and for different reasons; wherein the reasons at least include running and fighting; Acquire static facial images and dynamic facial images of cashmere goats under static and corresponding lighting conditions, and construct the facial images, static facial images, and dynamic facial images belonging to the same cashmere goat under the same lighting conditions into one piece of training data, thereby obtaining a training data set; The large model is fine-tuned using the training data set to obtain the first model.

[0046] In this embodiment, the training data in the lightweight training data set should include facial images of cashmere goats in motion, and these facial images are divided into different lighting conditions and different movement reasons, because lighting conditions and movement reasons will cause different characteristics of goat facial features. For example, when the light is dim and the tone is warm, the facial features of cashmere goats will show a special light and shadow effect, and the artificial light produced by lighting equipment at night will produce shadows under strong light sources on the faces of cashmere goats. When running, the face of the cashmere goat will shake and deform due to movement, which will enable the first model to learn the feature recognition ability under dynamic blur; during the fight, the facial muscles of the cashmere goat are tense, the position and posture of the horns may change, and the fur will also be messy due to intense movement, which will help the first model to grasp the facial features in the fight state.

[0047] In addition, the present invention also captures the static facial image of the cashmere goat in a static state (such as Figure 2 The first model is provided with a plurality of cashmere goat dynamic facial images, and the ...

[0048] The training process of the second model is similar to the above process, the only difference is that the training data is only obtained based on static facial images and dynamic facial images of cashmere goats in static state, which will not be repeated here.

[0049] Optionally, when the cause factor can be determined and judged not to meet the preset condition, it also includes: Calculate the number of the cause factors, and determine the waiting duration according to the number; wherein, when the number is larger, the set waiting duration is longer, and vice versa, the set waiting duration is shorter. After reaching the waiting duration, intercept a second set of consecutive video frames including the face information of each cashmere goat from the monitoring video data, and use the second model to perform identity recognition on the second set of consecutive video frames to obtain a second conclusion on whether the identity information of the corresponding cashmere goat belongs to the identity of twin lamb breeding.

[0050] In this embodiment, when the cause factors leading to the group movement of the flock can be determined, it is predicted that the flock will end the group movement state and turn into a stationary state in a short time. In this case, the second model can be called to perform sheep face recognition after the waiting duration. Since the second model is a lightweight model, the computing power load for performing sheep face recognition using it is lower. For the above waiting duration, the present invention is set to be determined according to the number of cause factors leading to the group movement of the flock. Specifically, when the number of the above cause factors is larger (for example, the breeder and the administrator enter the mixed breeding area successively), the time taken for all cause factors to disappear is generally longer or more uncertain. Correspondingly, the duration for which the flock maintains the group movement state is more uncertain, and at this time, the waiting duration is set to be longer; vice versa, when the number of the above cause factors is smaller, the time taken for all cause factors to disappear is generally shorter, and the probability that the flock quickly ends the group movement state is greater, and at this time, the waiting duration is set to be shorter.

[0051] Optionally, after reaching the waiting duration, intercepting a second set of consecutive video frames including the face information of each cashmere goat from the monitoring video data includes: Based on the monitoring video data of the cashmere goat mixed breeding area again, determine the latest group movement state information of the cashmere goats in the mixed breeding area, and judge whether the latest group movement state information meets the preset conditions; If not, intercept the second set of consecutive video frames including the face information of each cashmere goat from the monitoring video data.

[0052] In this embodiment, after reaching the waiting duration, the group movement state information of the flock is re-determined in the foregoing manner. If it is found that the flock is no longer in the group movement state, the second model can be called at this time to start the prediction analysis of the second set of consecutive video frames. If the flock is still in the group movement state, the first model still needs to analyze the identity of the cashmere goat based on the first set of consecutive video frames.

[0053] Refer to Figure 3As shown in the figure, an embodiment of the present invention further provides a cashmere goat identification and tracking system based on a large model. The system includes a processor and a memory, and the memory stores executable program code. The processor calls the executable program code stored in the memory to implement the following steps: Determine the group movement state information of the cashmere goats in the mixed breeding area based on the monitoring video data of the cashmere goat mixed breeding area, and determine whether the group movement state information meets the preset conditions; If the group movement state information meets the preset conditions, call the first model. The first model is constructed based on a large model and trained based on the facial images of cashmere goats in a dynamic state. Intercept the first set of consecutive video frames containing facial information of each cashmere goat from the monitoring video data, and use the first model to perform identity recognition on the first set of consecutive video frames to obtain the first conclusion on whether the identity information of the corresponding cashmere goat belongs to the identity of dual-lamb breeding; If the group movement state information does not meet the preset conditions, call the second model. The second model is constructed based on a non-large model and trained based on the static facial images and dynamic facial images of cashmere goats in a static state. Intercept the second set of consecutive video frames containing facial information of each cashmere goat from the monitoring video data, and use the second model to perform identity recognition on the second set of consecutive video frames to obtain the second conclusion on whether the identity information of the corresponding cashmere goat belongs to the identity of dual-lamb breeding.

[0054] An embodiment of the present invention further provides an electronic device, including: a memory storing executable program code; a processor coupled to the memory; the processor calls the executable program code stored in the memory to execute the method described in any one of the above embodiments.

[0055] An embodiment of the present invention further provides a computer storage medium, on which a computer program is stored. When the computer program is run by a processor, it executes the method described in any one of the above embodiments.

[0056] An embodiment of the present invention further provides a computer program product, which includes a computer program stored in a computer storage medium. When the computer program is executed by a processor of an electronic device, it implements the method described in any one of the above embodiments.

[0057] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the above-described system (if any) and device can refer to the corresponding processes in the foregoing method embodiments, and will not be described herein again.

[0058] In several embodiments provided by the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system or device, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of the devices or units can be in electrical, mechanical or other forms.

[0059] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0060] In addition, in each embodiment of the present application, the functional units can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

Claims

1. A cashmere goat identification and tracking method based on a large model, characterized in that: The method comprises the following steps: Determine the group movement state information of cashmere goats in the mixed-raising area based on the monitoring video data of the cashmere goats in the mixed-raising area, and judge whether the group movement state information meets the preset conditions; If the group motion state information meets the preset conditions, the first model is called, where the first model is constructed based on the large model and is trained based on the dynamic facial images of cashmere goats; A first group of continuous video frames containing facial information of each cashmere goat is intercepted from the monitoring video data, and the first model is used to perform identity recognition on the first group of continuous video frames to obtain a first conclusion of whether the identity information of the corresponding cashmere goat belongs to the twin lamb breeding identity; If the group motion state information does not meet the preset condition, calling the second model, the second model is constructed based on the non-large model and is trained based on the static facial image and the dynamic facial image of the cashmere goat in static state; A second group of continuous video frames containing facial information of each cashmere goat is intercepted from the monitoring video data, and the second model is used to perform identity recognition on the second group of continuous video frames to obtain a second conclusion on whether the identity information of the corresponding cashmere goat belongs to the twin lamb breeding identity.

2. The cashmere goat identification and tracking method based on a large model according to claim 1, characterized in that: Determining group motion state information of cashmere goats in the mixed-breeding area based on surveillance video data of the cashmere goats in the mixed-breeding area, and judging whether the group motion state information meets a preset condition, includes: Determine the group movement state information of cashmere goats in the mixed-breeding area based on the surveillance video data of the cashmere goat mixed-breeding area; When the group motion state information corresponds to a group motion state, further determining the cause factor causing the group motion state based on the monitoring video data; if the cause factor can be determined, determining that the preset condition is not satisfied; if the cause factor cannot be determined, determining that the preset condition is satisfied; When the group motion state information corresponds to a non-group motion state, it is determined that the preset condition is not satisfied.

3. A cashmere goat identification and tracking method based on a large model according to claim 2, characterized in that: The first model is trained in the following way: Acquire facial images of cashmere goats in motion under different lighting conditions and for different reasons; wherein the reasons at least include running and fighting; Acquire a static facial image and a dynamic facial image of a cashmere goat under static and corresponding lighting conditions, and construct the facial image, the static facial image, and the dynamic facial image under the same lighting conditions belonging to the same cashmere goat into a piece of training data, thereby obtaining a training data set; The large model is fine-tuned using the training data set to obtain the first model.

4. The cashmere goat identification and tracking method based on a large model according to claim 3 is characterized in that: When the cause factor can be determined and judged not to meet the preset conditions, it also includes: Calculating the number of the cause factors and determining the waiting time according to the number; wherein, when the number is greater, the waiting time is set longer, and vice versa, the waiting time is set shorter; After the waiting time is reached, a second group of continuous video frames containing facial information of each cashmere goat is intercepted from the monitoring video data, and the second model is used to perform identity recognition on the second group of continuous video frames to obtain a second conclusion on whether the identity information of the corresponding cashmere goat belongs to the twin lamb breeding identity.

5. The cashmere goat identification and tracking method based on a large model according to claim 4 is characterized in that: After the waiting time is reached, a second group of continuous video frames containing facial information of each cashmere goat is intercepted from the monitoring video data, including: Determine the latest group motion state information of the cashmere goats in the mixed-breeding area based on the surveillance video data of the cashmere goat mixed-breeding area again, and judge whether the latest group motion state information meets the preset conditions; If not, the second group of continuous video frames containing facial information of each cashmere goat is intercepted from the monitoring video data.

6. A cashmere goat identification and tracking system based on a large model, the system comprising a processor and a memory, the memory storing executable program code; characterized in that: The processor calls the executable program code stored in the memory to implement the following steps: Determine the group movement state information of cashmere goats in the mixed-raising area based on the monitoring video data of the cashmere goats in the mixed-raising area, and judge whether the group movement state information meets the preset conditions; If the group motion state information meets the preset conditions, the first model is called, where the first model is constructed based on the large model and is trained based on the dynamic facial images of cashmere goats; A first group of continuous video frames containing facial information of each cashmere goat is intercepted from the monitoring video data, and the first model is used to perform identity recognition on the first group of continuous video frames to obtain a first conclusion of whether the identity information of the corresponding cashmere goat belongs to the twin lamb breeding identity; If the group motion state information does not meet the preset condition, calling the second model, the second model is constructed based on the non-large model and is trained based on the static facial image and the dynamic facial image of the cashmere goat in static state; A second group of continuous video frames containing facial information of each cashmere goat is intercepted from the monitoring video data, and the second model is used to perform identity recognition on the second group of continuous video frames to obtain a second conclusion on whether the identity information of the corresponding cashmere goat belongs to the twin lamb breeding identity.

7. The cashmere goat identification and tracking system based on a large model according to claim 6 is characterized by: Determining group motion state information of cashmere goats in the mixed-breeding area based on surveillance video data of the cashmere goats in the mixed-breeding area, and judging whether the group motion state information meets a preset condition, includes: Determine the group movement state information of cashmere goats in the mixed-breeding area based on the surveillance video data of the cashmere goat mixed-breeding area; When the group motion state information corresponds to a group motion state, further determining the cause factor causing the group motion state based on the monitoring video data; if the cause factor can be determined, determining that the preset condition is not satisfied; if the cause factor cannot be determined, determining that the preset condition is satisfied; When the group motion state information corresponds to a non-group motion state, it is determined that the preset condition is not satisfied.

8. An electronic device comprising: A memory storing executable program code; A processor coupled to the memory; characterized in that: the processor calls the executable program code stored in the memory to execute the method according to any one of claims 1-5.

9. A computer storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 5 is executed.

10. A computer program product, comprising a computer program stored in a computer storage medium, characterized in that: When the computer program is executed by a processor of an electronic device, the method according to any one of claims 1 to 5 is implemented.

Images