Body condition scoring method and related device, electronic device and storage medium

Abstract

The present application discloses a body condition scoring method and related devices, electronic devices, and storage media. Among them, the body condition scoring method includes: performing a first evaluation on the first captured image of the object to be evaluated to obtain a first evaluation result of the object to be evaluated; wherein, the first captured image is obtained by capturing the object to be evaluated from a preset shooting perspective, and the first evaluation result includes the target score range of the object to be evaluated, and the target score range is selected from multiple preset score ranges; obtaining a second captured image of the object to be evaluated captured from the target shooting perspective; wherein, the target shooting perspective matches the target score range; performing a second evaluation on the second captured image based on the target evaluation strategy to obtain a second evaluation result of the object to be evaluated; wherein, the target evaluation strategy matches the target score range, and the second evaluation result includes the body condition score of the object to be evaluated. The above solution can improve the efficiency and accuracy of body condition scoring.

Description

Technical Field

[0001] This application relates to the field of image processing technologies, and particularly to a body condition scoring method, related devices, electronic devices, and storage media. Background Art

[0002] In scenarios related to breeding, such as beef cattle, dairy cattle breeding scenarios, etc., it is usually necessary to perform body condition scoring at different growth stages. Generally speaking, there is a most suitable body condition score corresponding to each growth stage, and by means such as feed feeding according to the growth stage, the body condition can be adjusted to the optimal score, which can not only maintain health, reduce the incidence of various metabolic diseases, but also greatly save feed, improve breeding efficiency, and reduce breeding costs.

[0003] Currently, body condition scoring generally relies on experienced technicians (such as breeding personnel, etc.). This method is highly dependent on technicians, and after a certain evaluation duration, affected by factors such as the working pressure of technicians, the evaluation efficiency and accuracy will inevitably decrease. In view of this, how to improve the efficiency and accuracy of body condition scoring has become an urgent problem to be solved. Summary of the Invention

[0004] The main technical problem to be solved by this application is to provide a body condition scoring method, related devices, electronic devices, and storage media, which can improve the efficiency and accuracy of body condition scoring.

[0005] To solve the above technical problem, in the first aspect of this application, a body condition scoring method is provided, including: performing a first evaluation on a first captured image of an object to be evaluated to obtain a first evaluation result of the object to be evaluated; wherein, the first captured image is captured from a preset shooting perspective of the object to be evaluated, the first evaluation result includes a target score range of the object to be evaluated, and the target score range is selected from multiple preset score ranges; obtaining a second captured image of the object to be evaluated captured from the target shooting perspective; wherein, the target shooting perspective matches the target score range; performing a second evaluation on the second captured image based on a target evaluation strategy to obtain a second evaluation result of the object to be evaluated; wherein, the target evaluation strategy matches the target score range, and the second evaluation result includes the body condition score of the object to be evaluated.

[0006] To solve the above technical problems, a second aspect of the present application provides a physical condition scoring device, including: a first evaluation module, an image acquisition module, and a second evaluation module. The first evaluation module is configured to perform a first evaluation on a first captured image of an object to be evaluated, and obtain a first evaluation result of the object to be evaluated. Among them, the first captured image is obtained by capturing the object to be evaluated from a preset shooting perspective, and the first evaluation result includes a target score range of the object to be evaluated, and the target score range is selected from multiple preset score ranges. The image acquisition module is configured to acquire a second captured image of the object to be evaluated captured from the target shooting perspective. Among them, the target shooting perspective matches the target score range. The second evaluation module is configured to perform a second evaluation on the second captured image based on a target evaluation strategy, and obtain a second evaluation result of the object to be evaluated. Among them, the target evaluation strategy matches the target score range, and the second evaluation result includes the physical condition score of the object to be evaluated.

[0007] To solve the above technical problems, a third aspect of the present application provides an electronic device, including a memory and a processor coupled to each other. The memory stores program instructions, and the processor is configured to execute the program instructions to implement the physical condition scoring method in the first aspect above.

[0008] To solve the above technical problems, a fourth aspect of the present application provides a computer-readable storage medium, storing program instructions that can be run by a processor, and the program instructions are used to implement the physical condition scoring method in the first aspect above.

[0009] In the above solution, a first evaluation is performed on the first captured image of the object to be evaluated to obtain a first evaluation result of the object to be evaluated. The first captured image is obtained by capturing the object to be evaluated from a preset shooting perspective. The first evaluation result includes a target score range of the object to be evaluated, and the target score range is selected from multiple preset score ranges. Based on this, a second captured image of the object to be evaluated captured from the target shooting perspective is obtained, and the target shooting perspective matches the target score range. Then, a second evaluation is performed on the second captured image based on the target evaluation strategy to obtain a second evaluation result of the object to be evaluated. The target evaluation strategy matches the target score range, and the second evaluation result includes the physical condition score of the object to be evaluated. On the one hand, since there is no need to rely on manual physical condition scoring, the evaluation efficiency can be improved, and the possibility of being interfered by subjective factors during the evaluation can be reduced. On the other hand, during the entire evaluation process, the physical condition score is determined in two stages from rough to fine. That is, in the first stage, a rough evaluation is performed based on the first captured image under the preset shooting perspective to determine the target score range. In the second stage, a fine evaluation is performed on the second captured image under the target shooting perspective that matches the target score range using the target evaluation strategy that matches the target score range, which is also beneficial to improving the evaluation accuracy. Therefore, the efficiency and accuracy of physical condition scoring are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 is a schematic flowchart of an embodiment of the body condition scoring method of the present application;

[0011] Figure 2 is a schematic diagram of an embodiment of a side view image;

[0012] Figure 3 is a schematic diagram of an embodiment of a top view image;

[0013] Figure 4 is a schematic diagram of another embodiment of a top view image;

[0014] Figure 5 is a schematic diagram of an embodiment of a rear view image;

[0015] Figure 6 is a schematic diagram of another embodiment of a side view image;

[0016] Figure 7 is a schematic framework diagram of an embodiment of the body condition scoring device of the present application;

[0017] Figure 8 is a schematic framework diagram of an embodiment of the electronic device of the present application;

[0018] Figure 9 is a schematic framework diagram of an embodiment of the computer-readable storage medium of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] The solutions of the embodiments of the present application will be described in detail below with reference to the accompanying drawings of the specification.

[0020] In the following description, specific details such as specific system architectures, interfaces, and technologies are presented for purposes of illustration rather than limitation, so as to thoroughly understand the present application.

[0021] The terms "system" and "network" are often used interchangeably herein. The term "and / or" herein merely describes an association relationship of associated objects and indicates that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " herein generally represents an "or" relationship between the associated objects before and after. Furthermore, "plurality" herein means two or more than two.

[0022] Please refer to Figure 1 , Figure 1 is a schematic flowchart of an embodiment of the body condition scoring method of the present application.

[0023] Specifically, the following steps may be included:

[0024] Step S11: Perform a first evaluation based on the first captured image of the object to be evaluated, and obtain the first evaluation result of the object to be evaluated.

[0025] In the embodiments of the present disclosure, the first captured image is obtained by capturing the object to be evaluated from a preset shooting perspective. Specifically, the preset shooting perspective may include, but is not limited to, side view, top view, etc., which are not limited herein. For example, when the preset shooting perspective includes a side view, the first captured image includes a side view image; or when the preset shooting perspective includes a top view, the first captured image includes a top view image, and so on for other cases, which will not be exemplified one by one here. Of course, the preset shooting perspective can also be set to other perspectives such as a rear view as needed, which is not limited herein.

[0026] In an implementation scenario, taking the preset shooting perspective including side view and top view as an example, the first evaluation can be performed only based on the side view image, or only based on the top view image, or also based on the side view image and the top view image, which is not limited herein. By performing the first evaluation in at least one of the side view and the top view in the above manner, it is beneficial to improve the accuracy of the rough evaluation in the first stage.

[0027] In an implementation scenario, the object to be evaluated can be set according to the actual application situation. For example, in the beef cattle breeding scenario, the object to be evaluated can be beef cattle; or in the dairy cattle breeding scenario, the object to be evaluated can be dairy cattle; or in the pig breeding scenario, the object to be evaluated can be pigs, and so on for other cases, which will not be exemplified one by one here.

[0028] In the embodiments of the present disclosure, the first evaluation result may include the target score range of the object to be evaluated, and the target score range is selected from multiple preset score ranges. Specifically, the multiple score ranges can also be set according to the actual application situation. Exemplarily, the multiple score ranges can be set to include a first score range and a second score range, and the preset limit value of the first score range is lower than the preset limit value of the second score range. It should be noted that the preset limit value can be a maximum value or a minimum value, which is not limited herein. For example, the multiple preset score ranges can include a first score range [1, 3] (i.e., 1 to 3 points) and a second score range (3, 5] (i.e., 3 to 5 points, excluding 3 points). Of course, the above example is only a possible situation in actual application, and does not limit the specific setting method of the multiple preset score ranges accordingly.

[0029] In an implementation scenario, the preset shooting angle can include a side view. The first captured image can include a side view image. Then, key point detection can be performed on the side view image to obtain a number of bone and joint points. The number of bone and joint points includes a target point and multiple reference points. Based on this, the angles between the lines connecting the multiple reference points to the target point can be obtained. And based on the magnitude relationship between the angle and the angle threshold, the first score interval or the second score interval can be selected from multiple preset score intervals as the target score interval. In the above manner, when the preset shooting angle includes a side view, by measuring the magnitude of the angle between the lines connecting the bone and joint points, the corresponding preset score interval is selected as the target score interval, which is beneficial to improving the accuracy of the target score interval.

[0030] In a specific implementation scenario, in order to improve the efficiency of bone and joint point detection, a bone and joint point detection model can be pre-trained. The bone and joint point detection model can include, but is not limited to, a convolutional neural network, etc. The network structure of the bone and joint point detection model is not limited here. Specifically, a number of sample images can be pre-collected, and the sample positions of the bone and joint points are marked in the sample images. Based on this, the key point detection can be performed on the sample images using the bone and joint point detection model to obtain the predicted positions of the bone and joint points, and based on the difference between the sample positions and the predicted positions, the network parameters of the key point detection model can be adjusted. It should be noted that for the specific measurement process of the difference, the technical details of loss functions such as mean square error can be referred to, and for the specific adjustment process of the parameters, the technical details of optimization methods such as gradient descent can be referred to, which will not be elaborated here.

[0031] In a specific implementation scenario, please refer to Figure 2 , Figure 2 is a schematic diagram of an embodiment of a side view image. As Figure 2 shown, taking dairy farming as an example, the number of bone and joint points can include, but is not limited to: clavicle point, hip joint point, and ischial joint point, etc. Then, the target point is the hip joint point, and the multiple reference points are the clavicle point and the ischial joint point. Then, the first line connecting the hip joint point and the clavicle point can be obtained, and the second line connecting the hip joint point and the ischial joint point can be obtained, and the angle between the first line and the second line can be obtained. For the convenience of description, the vector representing the first line can be denoted as and the vector representing the second line can be denoted as Then the angle θ can be expressed as Other situations can be deduced by analogy, and no further examples will be given here.

[0032] In a specific implementation scenario, when the included angle is greater than the angle threshold, the second score interval (e.g., (3, 5]) can be selected as the target score interval, while when the included angle is not greater than the angle threshold, the first score interval (e.g., [1, 3]) can be selected as the target score interval. In addition, the angle threshold can be specifically set according to the actual application situation. For example, the angle threshold can be set to 90 degrees, etc., which is not limited here.

[0033] In an implementation scenario, the preset shooting perspective can include a top view. The first captured image can include a top view image. Then, bone shape detection can be performed on the top view image to obtain the first shape detection results of several first reference bones. Based on the first shape detection results, the first score interval or the second score interval can be selected from multiple preset score intervals as the target score interval. In the above manner, when the preset shooting perspective includes a top view, by detecting the shapes of several first reference bones to select the corresponding preset score interval as the target score interval, it is beneficial to improve the accuracy of the target score interval.

[0034] In a specific implementation scenario, to improve the efficiency of bone shape detection, a bone shape detection model can be pre-trained. The bone shape detection model can include, but is not limited to, a convolutional neural network. The network structure of the bone shape detection model is not limited here. Specifically, several sample images can be pre-collected, and the sample shapes of several first reference bones are marked on the sample images. Based on this, the sample images can be detected using the bone shape detection model to obtain the predicted shapes of several first reference bones, and the network parameters of the bone shape detection model can be adjusted based on the difference between the sample shape and the predicted shape. It should be noted that for the specific measurement process of the difference, the technical details of loss functions such as cross-entropy can be referred to, and for the specific adjustment process of the parameters, the technical details of optimization methods such as gradient descent can be referred to, which will not be elaborated here.

[0035] In a specific implementation scenario, please refer to Figure 3 , Figure 3 which is a schematic diagram of an embodiment of a top view image. As Figure 3 shown, still taking dairy cow breeding as an example, the first reference bones can include the ischium and the clavicle, and the first shape detection results can include the shape of the ischium and the shape of the clavicle. The shapes can be divided into two types: rhombus and spherical. In addition, when both the ischium and the clavicle are rhombus-shaped, the first score interval (e.g., [1, 3]) can be selected as the target score interval, or when both the ischium and the clavicle are spherical-shaped, the second score interval (e.g., (3, 5]) can be selected as the target score interval. Other situations can be deduced by analogy, and no further examples will be given here.

[0036] It should be noted that the above-mentioned bone joint point detection model and bone shape detection model can be integrated into the body condition scoring model. That is, the body condition scoring model can include the bone joint point detection model and the bone shape detection model, and by judging whether the input image is a side view image or a top view image, the bone joint point detection model or the bone shape detection model is selected to perform a rough evaluation in the first stage.

[0037] Step S12: Obtain a second captured image of the object to be evaluated taken from the target shooting perspective.

[0038] In the embodiments of the present disclosure, the target shooting perspective matches the target score range. Exemplarily, as described above, the multiple preset score ranges at least include a first score range and a second score range, and the preset limit value of the first score range is lower than the preset limit value of the second score range. Then, the target shooting perspectives matching the first score range can include the rear view and the top view. Correspondingly, the second captured images can include rear view images and top view images. The target shooting perspectives matching the second score range can include the rear view and the side view. Correspondingly, the second captured images can include rear view images and side view images. Other situations can be deduced by analogy and will not be exemplified one by one here. In the above manner, on the basis of rough scoring, it is possible to further perform fine scoring by combining multi-perspective images, which is beneficial to improving the accuracy of body condition scoring compared with the scoring mechanism under a single perspective.

[0039] Step S13: Perform a second evaluation on the second captured image based on the target evaluation strategy to obtain a second evaluation result of the object to be evaluated.

[0040] In the embodiments of the present disclosure, the target evaluation strategy matches the target score range, and the second evaluation result includes the body condition score of the object to be evaluated.

[0041] In an implementation scenario, the target score range is the first score range, and the second captured image includes a rear view image and a top view image. Then, the shape of a number of second reference bones can be detected sequentially based on the rear view image to obtain the second shape detection results of the number of second reference bones. On this basis, in response to the second shape detection results meeting the first shape condition, the second evaluation result can be obtained based on the second shape detection results of the number of second reference bones. Alternatively, in response to the second shape detection results not meeting the first shape condition, the distance and shape of a number of third reference bones can be detected based on the top view image to obtain the second evaluation result, and the maximum physical condition score in the case of not meeting the first shape condition is not higher than the minimum physical condition score in the case of meeting the first shape condition. In the above manner, when the target score range is the relatively low first score range, by sequentially detecting the shapes of a number of second reference bones, and when the shape detection results do not meet the relevant conditions, continuing to detect the distance and shape of a number of third reference bones based on the top view image, and the maximum physical condition score in the case of not meeting the relevant conditions is not higher than the minimum physical condition score in the case of meeting the relevant conditions, it is possible to gradually narrow down the evaluation criteria and lock in the final physical condition score, which is beneficial to improving the accuracy of the second-stage fine evaluation when the first-stage rough evaluation is in the first score range.

[0042] In a specific implementation scenario, a number of second reference bones are arranged in sequence as: the clavicle, the ischium, and the ischial tuberosity. The first shape condition can include any one of the following: the clavicle is circular, the clavicle is rhomboid and the ischium is round, or both the clavicle and the ischium are rhomboid and the clavicle tubercle is round.

[0043] In a specific implementation scenario, to improve the efficiency of bone shape detection, a bone shape detection model can be pre-trained, so that the bone shape detection model can be used to detect the rear view image to obtain the second shape detection results of a number of second reference bones. It should be noted that the bone shape detection model and the bone shape detection model in the foregoing related description can be the same model. In this case, the sample image can be labeled with the sample shapes of a number of first reference bones and second reference bones. After detecting the sample image using the bone shape detection model, the predicted shapes of a number of first reference bones and second reference bones can be obtained, and the network parameters of the bone shape detection model can be adjusted based on the difference between the sample shape and the predicted shape of the first reference bone, and based on the difference between the sample shape and the predicted shape of the second reference bone. For the specific measurement process of the difference and the specific adjustment method of the parameters, reference can be made to the foregoing related description, which will not be elaborated here. Of course, the bone shape detection model can also be two different models from the bone shape detection model in the foregoing related description, which is not limited here.

[0044] In a specific implementation scenario, still taking the aforementioned second reference bone and the first shape condition as an example, when the collarbone is circular, the body condition score can be the first score value; when the collarbone is rhomboid and the ischium is round, the body condition score can be the second score value; when both the collarbone and the ischium are rhomboid and the collarbone tubercle is round, the body condition score can be the third score value. The first score value, the second score value, and the third score value gradually decrease and are all within the first score value range. Exemplarily, taking the first score value range as [1, 3] as an example, when the collarbone is circular, the body condition score can be 3; when the collarbone is rhomboid and the ischium is round, the body condition score can be 2.75; when both the collarbone and the ischium are rhomboid and the collarbone tubercle is round, the body condition score can be 2.5. Other cases can be deduced by analogy and will not be elaborated one by one here. In the above manner, a number of second reference bones are arranged in sequence as the collarbone, the ischium, and the ischium tubercle, and the corresponding first shape condition and the specific score values of the body condition score in different cases are set, which can gradually lock the body condition score through the constraint conditions and is beneficial to improving the accuracy of the body condition score.

[0045] In a specific implementation scenario, the third reference bone can at least include the short rib and the spine. Then, in response to the third shape detection result of the third reference bone not satisfying the second shape condition, the first distance from the outer end of the short rib to the spine is obtained, the second distance from the inner end of the short rib to the spine is obtained, and based on the distance ratio between the second distance and the first distance, the body condition score is obtained. The distance ratio is positively correlated with the body condition score, and the minimum body condition score under the condition of not satisfying the second shape condition is not lower than the score threshold. Please refer to Figure 4 , Figure 4 which is a schematic diagram of another embodiment of the top view image. As Figure 4 shown, still taking dairy cow breeding as an example, Figure 4 the white dotted line in [the figure] represents the spine, the solid line frame represents the short rib, the outermost side within the solid line frame is the outer side of the short rib (as shown by the white solid line), and the side of the solid line frame close to the white dotted line is the inner side of the short rib. For the convenience of description, the first distance can be denoted as D, and the second distance can be denoted as d. Other cases can be deduced by analogy and will not be elaborated one by one here. In addition, the larger the distance ratio (d / D) between the first distance D and the second distance d, the larger the body condition score; conversely, the smaller the distance ratio, the smaller the body condition score. Exemplarily, when the distance ratio is 0.5 (or within a numerical range centered on 0.5 such as 0.495 to 0.55), the body condition score can be 2.25; or when the distance ratio is 0.25 (or within a numerical range centered on 0.25 such as 0.245 to 0.255), the body condition score is 2.0. Other cases can be deduced by analogy and will not be elaborated one by one here.

[0046] In a specific implementation scenario, several third reference bones may further include the hip joint. The above second shape condition may include: abnormal protrusion of the spine, ribs, and hip joint. That is to say, when several third reference bones do not meet the second shape condition, the physical condition score can be determined by measuring the distance ratio, and the physical condition score at this time is not lower than the score threshold. On the contrary, when several third reference bones meet the second shape condition, it can be directly determined that the physical condition score at this time is lower than the score threshold. It should be noted that the score threshold can be set according to the actual application situation, such as it can be set to 2, which is not limited here. In addition, in order to improve the efficiency of determining whether the second shape condition is met, a binary classification model can be pre-trained. The binary classification model can include, but is not limited to, a convolutional neural network, and the network structure of the binary classification model is not limited here. Specifically, several sample images can be collected, and the sample images can be labeled with sample marks indicating whether the several third reference bones actually meet the second shape condition (for example, using sample mark 0 to indicate not meeting the second shape condition, and using sample mark 1 to indicate meeting the second shape condition). Based on this, the binary classification model can be used to detect the sample images to obtain prediction marks, and the prediction marks indicate whether the several third reference bones are predicted to meet the second shape condition. On this basis, the network parameters of the binary classification model can be adjusted based on the difference between the sample marks and the prediction marks. It should be noted that for the specific measurement process of the difference, the technical details of loss functions such as cross-entropy can be referred to, and for the specific adjustment method of the parameters, the technical details of optimization methods such as gradient descent can be referred to, which will not be elaborated here. In addition, the binary classification model can also be integrated into the physical condition scoring model, that is, the physical condition scoring model can further include the binary classification model in addition to the aforementioned bone joint point detection model and bone shape detection model.

[0047] In an implementation scenario, different from the foregoing situation, when the target score range is the second score range, the second captured image may include a rear view image and a side view image. Then, several reference ligaments can be detected based on the rear view image to obtain the first confidence detection results of the several reference ligaments. On this basis, in response to the first confidence detection results meeting the first confidence condition, a second evaluation result can be obtained based on the first confidence detection results of the several reference ligaments. Alternatively, in response to the first confidence detection results not meeting the first confidence condition, several fourth reference bones can be detected based on the side view image to obtain the second confidence detection results of the several fourth reference bones, and a second evaluation result can be obtained based on the second confidence detection results. Moreover, the minimum physical condition score when the first confidence condition is not met is not lower than the maximum physical condition score when the first confidence condition is met. In the above manner, when the target score range is the relatively high second score range, by performing the first confidence detection on several reference ligaments and directly obtaining the second evaluation result according to the first confidence detection results when the first confidence detection results meet the first confidence condition, and when the first confidence condition is not met, continuing to detect the side view image, and the minimum physical condition score when the first confidence condition is not met is not lower than the maximum physical condition score when the first confidence condition is met. Therefore, the evaluation criteria can be gradually narrowed down to lock in the final physical condition score, which is beneficial to improving the accuracy of the second-stage fine evaluation when the first-stage rough evaluation is in the second score range.

[0048] In a specific implementation scenario, the several reference ligaments may include the sacral ligament and the root of tail ligament, and the first confidence detection results may include the first confidence scores of the sacral ligament and the root of tail ligament. Please refer to Figure 5 , Figure 5 is a schematic diagram of an embodiment of the rear view image. As Figure 5 shown, still taking the dairy cow breeding scenario as an example, the sacral ligament is located between the spine and the collarbone, and the root of tail ligament is located between the ischium and the tailbone. Other situations can be deduced by analogy and will not be elaborated one by one here.

[0049] In a specific implementation scenario, to improve the detection efficiency of the reference ligament, a ligament detection model can be pre-trained. The ligament detection model can include, but is not limited to, a convolutional neural network, etc. The network parameters of the ligament detection model are not limited herein. Specifically, a number of sample images can be pre-collected, and the sample regions of the reference ligament are labeled on the sample images. On this basis, the ligament detection model can be used to detect the sample images to obtain the predicted regions and prediction confidence levels of the reference ligament, and the network parameters of the ligament detection model can be adjusted based on the differences between the sample regions and the predicted regions. It should be noted that for the measurement process of the differences, the technical details of loss functions such as cross-entropy can be referred to, and for the adjustment method of the parameters, the technical details of optimization methods such as gradient descent can be referred to, which will not be elaborated herein. In addition, the ligament detection model can be integrated into the body condition scoring model. For the specific composition of the body condition scoring model, the relevant descriptions above can be referred to, which will not be elaborated herein.

[0050] In a specific implementation scenario, the first confidence condition can be set to include any one of the following: the first confidence scores of both the sacral ligament and the coccygeal ligament are higher than the first confidence threshold, the first confidence score of the sacral ligament is higher than the first confidence threshold and the first confidence score of the coccygeal ligament is lower than the second confidence score, the first confidence score of the sacral ligament is lower than the second confidence score and the coccygeal ligament is not detected. It should be noted that the first confidence threshold is not lower than the second confidence threshold, and the first confidence threshold and the second confidence threshold can be set according to the actual application situation. For example, the first confidence threshold can be set to the same value as the second confidence threshold. Of course, the first confidence threshold can also be set to a different value from the second confidence threshold, which is not limited herein.

[0051] In a specific implementation scenario, when the first confidence scores of both the sacral ligament and the coccygeal ligament are higher than the first confidence threshold (i.e., high-confidence sacral ligament and coccygeal ligament are detected), the body condition score is the fourth score; when the first confidence score of the sacral ligament is higher than the first confidence threshold and the first confidence score of the coccygeal ligament is lower than the second confidence score (i.e., high-confidence sacral ligament and low-confidence coccygeal ligament are detected), the body condition score is the fifth score; when the first confidence score of the sacral ligament is lower than the second confidence score and the coccygeal ligament is not detected (i.e., low-confidence sacral ligament and no coccygeal ligament are detected), the body condition score is the sixth score. The fourth score, the fifth score, and the sixth score gradually increase and are all within the second score interval. Exemplarily, the fourth score can be set to 3.25, the fifth score can be 3.5, and the sixth score can be 3.75.

[0052] In a specific implementation scenario, in the case where neither the sacral ligament nor the caudal root ligament is detected, the detection of lateral view images (such as left view images and right view images) can continue. Specifically, a number of fourth reference bones can include the ends of the short ribs and the hip joints, and the second confidence detection result includes the second confidence scores of the ends of the short ribs and the hip joints, that is, at this time, mainly detect whether the ends of the short ribs and the hip joints are visible. Please refer to Figure 6 , Figure 6 which is a schematic diagram of an embodiment of a lateral view image. As Figure 6 shown, still taking the dairy farming scenario as an example, the end of the short rib is circled by the left solid line box, and the hip joint is circled by the right solid line box. Other situations can be deduced by analogy, and no further examples will be given here.

[0053] In a specific implementation scenario, in order to improve the efficiency of bone detection, a bone and joint detection model can be pre-trained. The bone and joint detection model can include, but is not limited to, a convolutional neural network, and the network structure of the bone and joint detection model is not limited here. It should be noted that the functions respectively implemented by the bone and joint detection model and the aforementioned ligament detection model can be completed by an integrated detection model, that is, the integrated detection model can detect the above-mentioned several reference ligaments and several fourth reference bones, and obtain the first confidence detection results of the above-mentioned several reference ligaments and the second confidence detection results of the above-mentioned several fourth reference bones. Of course, the bone and joint detection model can also be two independent network models from the aforementioned ligament detection model, which is not limited here.

[0054] In a specific implementation scenario, still taking several fourth reference bones including the ends of short ribs and the hip joint as an example, when the second confidence scores of both the ends of short ribs and the hip joint are higher than the third confidence threshold (i.e., high-confidence ends of short ribs and hip joint are detected), the body condition score is the seventh score; when the second confidence score of the hip joint is higher than the third confidence threshold and the second confidence score of the ends of short ribs is lower than the fourth confidence threshold (i.e., high-confidence hip joint and low-confidence ends of short ribs are detected), the body condition score is the eighth score; when the second confidence score of the hip joint is lower than the fourth confidence threshold and the ends of short ribs are not detected (i.e., low-confidence hip joint and no ends of short ribs are detected), the body condition score is the ninth score; when neither the ends of short ribs nor the hip joint are detected, the body condition score is the tenth score. The seventh score, the eighth score, the ninth score, and the tenth score gradually increase and are all within the second score interval. It should be noted that the third confidence threshold is not lower than the fourth confidence threshold, and the third confidence threshold and the fourth confidence threshold can be set according to the actual application situation. For example, the third confidence threshold can be set to the same value as the fourth confidence threshold. Of course, the third confidence threshold can also be set to a different value from the fourth confidence threshold, which is not limited here. Exemplarily, the seventh score can be set to 4, the eighth score can be set to 4.25, the ninth score can be set to 4.5, and the tenth score can be set to 5.

[0055] In the above solution, a first evaluation is performed on the first captured image of the object to be evaluated to obtain the first evaluation result of the object to be evaluated. The first captured image is obtained by capturing the object to be evaluated from a preset shooting perspective. The first evaluation result includes the target score interval of the object to be evaluated, and the target score interval is selected from multiple preset score intervals. Based on this, a second captured image obtained by capturing the object to be evaluated from the target shooting perspective is further obtained, and the target shooting perspective matches the target score interval. Then, a second evaluation is performed on the second captured image based on the target evaluation strategy, and the target evaluation strategy matches the target score interval. The second evaluation result includes the body condition score of the object to be evaluated. On the one hand, since there is no need to rely on manual body condition scoring, the evaluation efficiency can be improved, and the possibility of being interfered by subjective factors in the evaluation can be reduced. On the other hand, during the entire evaluation process, the body condition score is determined in two stages from coarse to fine. That is, in the first stage, a rough evaluation is performed based on the first captured image under the preset shooting perspective to determine the target score interval. In the second stage, a fine evaluation is performed on the second captured image under the target shooting perspective that matches the target score interval using the target evaluation strategy that matches the target score interval, which is also beneficial to improving the evaluation accuracy. Therefore, the efficiency and accuracy of the body condition score are improved.

[0056] Please refer to Figure 7 , Figure 7It is a framework schematic diagram of an embodiment of the physical condition scoring device 70 of the present application. The physical condition scoring device 70 includes: a first evaluation module 71, an image acquisition module 72, and a second evaluation module 73. The first evaluation module 71 is configured to perform a first evaluation on the first captured image of the object to be evaluated to obtain a first evaluation result of the object to be evaluated. Among them, the first captured image is obtained by capturing the object to be evaluated from a preset shooting perspective. The first evaluation result includes a target score range of the object to be evaluated, and the target score range is selected from multiple preset score ranges. The image acquisition module 72 is configured to acquire a second captured image of the object to be evaluated captured from the target shooting perspective. Among them, the target shooting perspective matches the target score range. The second evaluation module 73 is configured to perform a second evaluation on the second captured image based on the target evaluation strategy to obtain a second evaluation result of the object to be evaluated. Among them, the target evaluation strategy matches the target score range, and the second evaluation result includes the physical condition score of the object to be evaluated.

[0057] In the above solution, on the one hand, since there is no need to rely on manual physical condition scoring anymore, the evaluation efficiency can be improved, and the possibility of being interfered by subjective factors during the evaluation can be reduced. On the other hand, during the entire evaluation process, the physical condition score is determined from coarse to fine in two stages. That is, in the first stage, a rough evaluation is performed based on the first captured image under the preset shooting perspective to determine the target score range. In the second stage, a fine evaluation is performed on the second captured image under the target shooting perspective that matches the target score range by using the target evaluation strategy that matches the target score range. This is also beneficial to improving the evaluation accuracy. Therefore, the efficiency and accuracy of the physical condition scoring are improved.

[0058] In some disclosed embodiments, the preset shooting perspective includes at least one of a side view and a top view. And when the preset shooting perspective includes a side view, the first captured image includes a side view image. When the preset shooting perspective includes a top view, the first captured image includes a top view image.

[0059] Therefore, by performing the first evaluation from at least one of the side view and the top view, it is beneficial to improve the accuracy of the rough evaluation in the first stage.

[0060] In some disclosed embodiments, the preset shooting perspective includes a side view, the first captured image includes a side view image, and the first evaluation module 71 includes a key point detection sub-module for performing key point detection on the side view image to obtain a plurality of bone and joint points; wherein, the plurality of bone and joint points include a target point and a plurality of reference points; the first evaluation module 71 includes an included angle acquisition sub-module for acquiring the included angles between the connecting lines of the plurality of reference points and the target point respectively; the first evaluation module 71 includes a first selection sub-module for selecting a first score interval or a second score interval from a plurality of preset score intervals as the target score interval based on the size relationship between the included angle and the angle threshold; wherein, the plurality of preset score intervals include a first score interval and a second score interval, and the preset limit value of the first score interval is lower than the preset limit value of the second score interval.

[0061] Therefore, when the preset shooting perspective includes a side view, by measuring the size of the included angle between the connecting lines of the bone and joint points, and selecting the corresponding preset score interval as the target score interval, it is beneficial to improve the accuracy of the target score interval.

[0062] In some disclosed embodiments, the preset shooting perspective includes a top view, the first captured image includes a top view image, and the first evaluation module 71 includes a first detection sub-module for performing bone shape detection on the top view image to obtain the first shape detection results of a plurality of first reference bones; the first evaluation module 71 includes a second selection sub-module for selecting a first score interval or a second score interval from a plurality of preset score intervals as the target score interval based on the first shape detection results; wherein, the plurality of preset score intervals include a first score interval and a second score interval, and the preset limit value of the first score interval is lower than the preset limit value of the second score interval.

[0063] Therefore, when the preset shooting perspective includes a top view, by detecting the shapes of a plurality of first reference bones and selecting the corresponding preset score interval as the target score interval, it is beneficial to improve the accuracy of the target score interval.

[0064] In some disclosed embodiments, the plurality of preset score intervals include at least a first score interval and a second score interval, and the preset limit value of the first score interval is lower than the preset limit value of the second score interval; wherein, the target shooting perspectives matching the first score interval include a rear view and a top view, and the target shooting perspectives matching the second score interval include a rear view and a side view.

[0065] Therefore, on the basis of rough scoring, it is possible to further perform fine scoring by combining multi-perspective images, which is beneficial to improving the accuracy of the body condition score compared with the scoring mechanism under a single perspective.

[0066] In some disclosed embodiments, the target score range is the first score range. The second captured image includes a rear view image and a top view image. The second evaluation module 73 includes a second detection sub-module for sequentially performing shape detection on a plurality of second reference bones based on the rear view image to obtain second shape detection results of the plurality of second reference bones; the second evaluation module 73 includes a first shape evaluation sub-module for obtaining a second evaluation result based on the second shape detection results of the plurality of second reference bones in response to the second shape detection results satisfying the first shape condition; the second evaluation module 73 includes a second shape evaluation sub-module for obtaining a second evaluation result by performing spacing and shape detection on a plurality of third reference bones based on the top view image in response to the second shape detection results not satisfying the first shape condition; wherein, the maximum physical condition score in the case of not satisfying the first shape condition is not higher than the minimum physical condition score in the case of satisfying the first shape condition.

[0067] Therefore, when the target score range is the relatively low first score range, by sequentially performing shape detection on a plurality of second reference bones and continuing to perform spacing and shape detection on a plurality of third reference bones based on the top view image when the shape detection results do not satisfy the relevant conditions, and the maximum physical condition score in the case of not satisfying the relevant conditions is not higher than the minimum physical condition score in the case of satisfying the relevant conditions, it is possible to gradually narrow down the evaluation criteria and lock in the final physical condition score, which is beneficial to improving the accuracy of the second-stage fine evaluation when the first-stage rough evaluation is in the first score range.

[0068] In some disclosed embodiments, the plurality of second reference bones are sequentially arranged as: collarbone, ischium, ischial tuberosity; wherein, the first shape condition includes any one of the following: the collarbone is circular, the collarbone is rhomboid and the ischium is round, the collarbone and the ischium are both rhomboid and the collarbone tubercle is round, and in the case where the collarbone is circular, the physical condition score is the first score, in the case where the collarbone is rhomboid and the ischium is round, the physical condition score is the second score, in the case where the collarbone and the ischium are both rhomboid and the collarbone tubercle is round, the physical condition score is the third score, and the first score, the second score and the third score gradually decrease and are all within the first score range.

[0069] Therefore, arranging the plurality of second reference bones sequentially as the collarbone, the ischium, and the ischial tuberosity, and setting the corresponding first shape condition and the specific scores of the physical condition score in different cases can gradually lock in the physical condition score through the constraint conditions, which is beneficial to improving the accuracy of the physical condition score.

[0070] In some disclosed embodiments, the third reference bone includes at least the short rib and the backbone. The second shape evaluation sub-module includes a distance acquisition unit, which is configured to, in response to the third shape detection result of the third reference bone not meeting the second shape condition, acquire a first distance from the outer end of the short rib to the backbone and a second distance from the inner end of the short rib to the backbone; the second shape evaluation sub-module includes a physical condition scoring unit, which is configured to obtain a physical condition score based on the distance ratio between the second distance and the first distance; wherein the distance ratio is positively correlated with the physical condition score, and the minimum physical condition score under the condition of not meeting the second shape condition is not lower than the score threshold.

[0071] Therefore, when the third shape detection result of the third reference bone does not meet the second shape condition, using the distance ratio to perform the physical condition scoring is beneficial to improving the accuracy of the physical condition scoring.

[0072] In some disclosed embodiments, the second shape evaluation sub-module includes a score determination unit, which is configured to, in response to the third shape detection result of the third reference bone meeting the second shape condition, determine that the physical condition score is lower than the score threshold.

[0073] Therefore, when the third shape detection result of the third reference bone meets the second shape condition, directly determining that the physical condition score is lower than the score threshold is beneficial to improving the speed of the physical condition scoring.

[0074] In some disclosed embodiments, the target score range is the second score range. The second captured image includes a rear view image and a side view image. The second evaluation module 73 includes a ligament detection sub-module, which is configured to detect a plurality of reference ligaments based on the rear view image to obtain a first confidence detection result of the plurality of reference ligaments; the second evaluation module 73 includes a first confidence evaluation sub-module, which is configured to, in response to the first confidence detection result meeting the first confidence condition, obtain a second evaluation result based on the first confidence detection result of the plurality of reference ligaments; the second evaluation module 73 includes a second confidence evaluation sub-module, which is configured to, in response to the first confidence detection result not meeting the first confidence condition, detect a plurality of fourth reference bones based on the side view image to obtain a second confidence detection result of the plurality of fourth reference bones, and obtain a second evaluation result based on the second confidence detection result; wherein the minimum physical condition score under the condition of not meeting the first confidence condition is not lower than the maximum physical condition score under the condition of meeting the first confidence condition.

[0075] Therefore, when the target score range is the relatively high second score range, by performing a first confidence detection on a number of reference ligaments and directly obtaining a second evaluation result based on the first confidence detection result when the first confidence detection result meets the first confidence condition, and when the first confidence condition is not met, continuing to detect the lateral view image, and the minimum body condition score under the condition that the first confidence condition is not met is not lower than the maximum body condition score under the condition that the first confidence condition is met, it is possible to gradually narrow down the evaluation criteria and lock in the final body condition score, which is beneficial to improving the accuracy of the second-stage fine evaluation when the first-stage rough evaluation is in the second score range.

[0076] In some disclosed embodiments, the number of reference ligaments includes the sacral ligament and the root of the tail ligament, and the first confidence detection result includes the first confidence scores of the sacral ligament and the root of the tail ligament; wherein, the first confidence condition includes any one of the following: the first confidence scores of the sacral ligament and the root of the tail ligament are both higher than the first confidence threshold, the first confidence score of the sacral ligament is higher than the first confidence threshold and the first confidence score of the root of the tail ligament is lower than the second confidence score, the first confidence score of the sacral ligament is lower than the second confidence score and the root of the tail ligament is not detected, and when the first confidence scores of the sacral ligament and the root of the tail ligament are both higher than the first confidence threshold, the body condition score is the fourth score, when the first confidence score of the sacral ligament is higher than the first confidence threshold and the first confidence score of the root of the tail ligament is lower than the second confidence score, the body condition score is the fifth score, when the first confidence score of the sacral ligament is lower than the second confidence score and the root of the tail ligament is not detected, the body condition score is the sixth score, and the fourth score, the fifth score and the sixth score gradually increase and are all within the second score range.

[0077] Therefore, setting the number of reference ligaments to include the sacral ligament and the root of the tail ligament, setting the first confidence detection result to include the first confidence scores of the sacral ligament and the root of the tail ligament, and corresponding different body condition scores based on different situations of the first confidence scores is beneficial to gradually locking in the body condition score by the constraint conditions and is beneficial to improving the accuracy of the body condition score.

[0078] In some disclosed embodiments, the number of fourth reference bones includes the ends of the short ribs and the hip joints, and the second confidence detection result includes the second confidence scores of the ends of the short ribs and the hip joints; wherein, when the second confidence scores of the ends of the short ribs and the hip joints are both higher than the third confidence threshold, the body condition score is the seventh score, when the second confidence score of the hip joint is higher than the third confidence threshold and the second confidence score of the ends of the short ribs is lower than the fourth confidence threshold, the body condition score is the eighth score, when the second confidence score of the hip joint is lower than the fourth confidence threshold and the ends of the short ribs are not detected, the body condition score is the ninth score, and when neither the ends of the short ribs nor the hip joints are detected, the body condition score is the tenth score, and the seventh score, the eighth score, the ninth score and the tenth score gradually increase and are all within the second score range.

[0079] Therefore, by setting a number of fourth reference bones to include the ends of the short ribs and the hip joints, setting the second confidence detection result to include the second confidence score of the ends of the short ribs and the hip joints, and corresponding different body condition scores to different situations of the second confidence score, it is beneficial to gradually lock the body condition score by the constraint conditions and improve the accuracy of the body condition score.

[0080] Please refer to Figure 8 , Figure 8 which is a schematic diagram of the framework of an embodiment of the electronic device 80 of the present application. The electronic device 80 includes a memory 81 and a processor 82 that are coupled to each other. Program instructions are stored in the memory 81, and the processor 82 is configured to execute the program instructions to implement the steps in any of the above-described embodiments of the body condition scoring method. Specifically, the electronic device 80 may include, but is not limited to: a desktop computer, a laptop computer, a server, a mobile phone, a tablet computer, etc., which are not limited herein.

[0081] Specifically, the processor 82 is configured to control itself and the memory 81 to implement the steps in any of the above-described embodiments of the body condition scoring method. The processor 82 may also be referred to as a CPU (Central Processing Unit). The processor 82 may be an integrated circuit chip with signal processing capabilities. The processor 82 may also be a general-purpose processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. Additionally, the processor 82 may be implemented jointly by integrated circuit chips.

[0082] For the above solution, on the one hand, since there is no need to rely on manual body condition scoring anymore, the evaluation efficiency can be improved, and the possibility of being interfered by subjective factors during the evaluation can be reduced. On the other hand, during the entire evaluation process, the body condition score is determined from rough to fine in two stages. That is, in the first stage, a rough evaluation is performed based on the first captured image under the preset shooting angle to determine the target score range. In the second stage, a fine evaluation is performed on the second captured image under the target shooting angle that matches the target score range using the target evaluation strategy that matches the target score range, which is also beneficial to improving the evaluation accuracy. Therefore, the efficiency and accuracy of the body condition score are improved.

[0083] Please refer to Figure 9 , Figure 9It is a schematic framework diagram of an embodiment of the computer-readable storage medium 90 of the present application. The computer-readable storage medium 90 stores program instructions 91 that can be run by a processor, and the program instructions 91 are used to implement the steps in any of the above-described embodiment of the physical condition scoring method.

[0084] In the above solution, on the one hand, since there is no need to rely on manual physical condition scoring anymore, the evaluation efficiency can be improved, and the possibility of being interfered by subjective factors during the evaluation can be reduced. On the other hand, during the entire evaluation process, the physical condition score is determined from rough to fine in two stages. That is, in the first stage, a rough evaluation is performed based on the first captured image under a preset shooting angle to determine the target score range. In the second stage, a fine evaluation is performed on the second captured image under the target shooting angle that matches the target score range by using the target evaluation strategy that matches the target score range, which is also beneficial to improving the evaluation accuracy. Therefore, the efficiency and accuracy of the physical condition scoring are improved.

[0085] In some embodiments, the functions or modules included in the device provided by the present disclosure can be used to execute the methods described in the above method embodiments. The specific implementation can refer to the description of the above method embodiments. For the sake of brevity, it will not be repeated here.

[0086] The above descriptions of the various embodiments tend to emphasize the differences between the various embodiments. The same or similar parts can be referred to each other. For the sake of brevity, they will not be repeated in this article.

[0087] In the several embodiments provided by the present application, it should be understood that the disclosed methods and devices can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of modules or 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 some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection to each other can be through some interfaces. The indirect coupling or communication connection of the device or unit can be in electrical, mechanical or other forms.

[0088] 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 can be located in one place, or they can 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.

[0089] In addition, each functional unit in various embodiments of the present application may be integrated into one processing unit, may exist separately physically for each unit, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of a software functional unit.

[0090] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it may be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, may be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor to execute all or part of the steps of the methods in various embodiments of the present application. The foregoing storage medium includes: various media such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc that can store program codes.

Claims

1. A body condition scoring method, characterized in that Including: Performing a first evaluation on a first captured image of an object to be evaluated to obtain a first evaluation result of the object to be evaluated; wherein, the first captured image is obtained by capturing the object to be evaluated from a preset shooting perspective, the preset shooting perspective includes at least one of a side view and a top view, the first evaluation result includes a target score range of the object to be evaluated, and the target score range is selected from a plurality of preset score ranges, the plurality of preset score ranges include a first score range and a second score range, and the preset limit value of the first score range is lower than the preset limit value of the second score range; Obtaining a second captured image of the object to be evaluated captured from a target shooting perspective; wherein, the target shooting perspective matches the target score range, and the target shooting perspectives matching the first score range include a rear view and a top view, and the target shooting perspectives matching the second score range include a rear view and a side view; Performing a second evaluation on the second captured image based on a target evaluation strategy to obtain a second evaluation result of the object to be evaluated; wherein, the target evaluation strategy matches the target score range, and the second evaluation result includes a physical condition score of the object to be evaluated.

2. The method according to claim 1, characterized in that, When the preset shooting perspective includes the side view, the first captured image includes a side view image, and when the preset shooting perspective includes the top view, the first captured image includes a top view image.

3. The method according to claim 2, wherein The preset shooting perspective includes the side view, the first captured image includes the side view image, and performing a first evaluation on the first captured image of the object to be evaluated to obtain a first evaluation result of the object to be evaluated includes: Performing key point detection on the side view image to obtain a number of bone joint points; wherein, the number of bone joint points includes a target point and a plurality of reference points; Obtaining the included angles between the connections of the plurality of reference points and the target point respectively; Based on the magnitude relationship between the included angle and an angle threshold, selecting the first score range or the second score range from the plurality of preset score ranges as the target score range.

4. The method according to claim 2, wherein The preset shooting perspective includes the top view, the first captured image includes the top view image, and performing a first evaluation on the first captured image of the object to be evaluated to obtain a first evaluation result of the object to be evaluated includes: Performing bone shape detection on the top view image to obtain a first shape detection result of a number of first reference bones; Based on the first shape detection result, selecting the first score range or the second score range from the plurality of preset score ranges as the target score range.

5. The method according to claim 1, characterized in that, The target score range is the first score range, the second captured image includes a rear view image and a top view image, and performing a second evaluation on the second captured image based on a target evaluation strategy to obtain a second evaluation result of the object to be evaluated includes: Sequentially performing shape detection on a number of second reference bones based on the rear view image to obtain a second shape detection result of the number of second reference bones; In response to the second shape detection result satisfying the first shape condition, based on the second shape detection results of the plurality of second reference bones, the second evaluation result is obtained; wherein, the plurality of second reference bones are arranged in sequence as: clavicle, ischium, ischial tuberosity, and the first shape condition includes any one of the following: the clavicle is circular, the clavicle is rhomboid and the ischium is round, the clavicle and the ischium are both rhomboid and the clavicle tuberosity is round; In response to the second shape detection result not satisfying the first shape condition, the distance and shape of a plurality of third reference bones are detected based on the top view image to obtain the second evaluation result; Among them, the maximum physical condition score in the case of not satisfying the first shape condition is not higher than the minimum physical condition score in the case of satisfying the first shape condition.

6. The method according to claim 5, wherein In the case where the clavicle is circular, the physical condition score is the first score value, in the case where the clavicle is rhomboid and the ischium is round, the physical condition score is the second score value, and in the case where the clavicle and the ischium are both rhomboid and the clavicle tuberosity is round, the physical condition score is the third score value, and the first score value, the second score value, and the third score value gradually decrease and are all within the first score value range.

7. The method according to claim 5, characterized in that, The third reference bones at least include short ribs and the spine, and the distance and shape of a plurality of third reference bones are detected based on the top view image to obtain the second evaluation result, including: In response to the third shape detection result of the third reference bone not satisfying the second shape condition, the first distance from the outer end of the short rib to the spine is obtained, and the second distance from the inner end of the short rib to the spine is obtained; wherein, the third reference bone further includes the hip joint, and the second shape condition includes: the spine, the ribs, and the hip joint are abnormally prominent; Based on the distance ratio between the second distance and the first distance, the physical condition score is obtained; Among them, the distance ratio is positively correlated with the physical condition score, and the minimum physical condition score in the case of not satisfying the second shape condition is not lower than the score threshold.

8. The method according to claim 7, characterized in that, The method further includes: In response to the third shape detection result of the third reference bone satisfying the second shape condition, it is determined that the physical condition score is lower than the score threshold.

9. The method according to claim 4, wherein The target score range is the second score range, and the second captured image includes a rear view image and a side view image. The second evaluation of the second captured image based on the target evaluation strategy to obtain the second evaluation result of the object to be evaluated includes: Detecting a plurality of reference ligaments based on the rear view image to obtain the first confidence detection result of the plurality of reference ligaments; In response to the first confidence detection result satisfying the first confidence condition, based on the first confidence detection results of the several reference ligaments, the second evaluation result is obtained; wherein, the several reference ligaments include the sacral ligament and the root of tail ligament, the first confidence detection results include the first confidence scores of the sacral ligament and the root of tail ligament, and the first confidence condition includes any one of the following: the first confidence scores of both the sacral ligament and the root of tail ligament are higher than the first confidence threshold, the first confidence score of the sacral ligament is higher than the first confidence threshold and the first confidence score of the root of tail ligament is lower than the second confidence score, the first confidence score of the sacral ligament is lower than the second confidence score and the root of tail ligament is not detected; In response to the first confidence detection result not satisfying the first confidence condition, several fourth reference bones are detected based on the lateral view image to obtain the second confidence detection results of the several fourth reference bones, and the second evaluation result is obtained based on the second confidence detection results; Wherein, the minimum body condition score under the condition of not satisfying the first confidence condition is not lower than the maximum body condition score under the condition of satisfying the first confidence condition.

10. The method according to claim 9, characterized in that, When the first confidence scores of both the sacral ligament and the root of tail ligament are higher than the first confidence threshold, the body condition score is the fourth score; when the first confidence score of the sacral ligament is higher than the first confidence threshold and the first confidence score of the root of tail ligament is lower than the second confidence score, the body condition score is the fifth score; when the first confidence score of the sacral ligament is lower than the second confidence score and the root of tail ligament is not detected, the body condition score is the sixth score; the fourth score, the fifth score and the sixth score increase gradually and are all within the second score interval.

11. The method according to claim 9, wherein The several fourth reference bones include the ends of short ribs and the hip joints, and the second confidence detection results include the second confidence scores of the ends of short ribs and the hip joints; Wherein, when the second confidence scores of both the ends of short ribs and the hip joints are higher than the third confidence threshold, the body condition score is the seventh score; when the second confidence score of the hip joint is higher than the third confidence threshold and the second confidence score of the ends of short ribs is lower than the fourth confidence threshold, the body condition score is the eighth score; when the second confidence score of the hip joint is lower than the fourth confidence threshold and the ends of short ribs are not detected, the body condition score is the ninth score; when neither the ends of short ribs nor the hip joints are detected, the body condition score is the tenth score; the seventh score, the eighth score, the ninth score and the tenth score increase gradually and are all within the second score interval.

12. A body condition scoring device, characterized in that, Comprising: The first evaluation module is configured to perform a first evaluation based on a first captured image of the object to be evaluated, so as to obtain a first evaluation result of the object to be evaluated; wherein, the first captured image is obtained by capturing the object to be evaluated from a preset shooting perspective, the preset shooting perspective includes at least one of a side view and a top view, the first evaluation result includes a target score range of the object to be evaluated, and the target score range is selected from a plurality of preset score ranges, the plurality of preset score ranges include a first score range and a second score range, and a preset limit value of the first score range is lower than a preset limit value of the second score range; The image acquisition module is configured to acquire a second captured image of the object to be evaluated captured from a target shooting perspective; wherein, the target shooting perspective matches the target score range, the target shooting perspectives matching the first score range include a rear view and a top view, and the target shooting perspectives matching the second score range include a rear view and a side view; The second evaluation module is configured to perform a second evaluation on the second captured image based on a target evaluation strategy, so as to obtain a second evaluation result of the object to be evaluated; wherein, the target evaluation strategy matches the target score range, and the second evaluation result includes a physical condition score of the object to be evaluated.

13. An electronic device, characterized in that, It includes a memory and a processor which are mutually coupled, program instructions are stored in the memory, and the processor is configured to execute the program instructions to implement the physical condition scoring method according to any one of claims 1 to 11.

14. A computer-readable storage medium, characterized in that, Program instructions capable of being run by a processor are stored, and the program instructions are used to implement the physical condition scoring method according to any one of claims 1 to 11.

Images