Intelligent poultry auxiliary feeding method and system

Abstract

The invention provides an intelligent poultry auxiliary feeding system and method. The system comprises a server and a poultry diagnosis device. The server comprises a poultry image recognition and diagnosis artificial intelligence model. The poultry diagnosis device is used for being connected with a server through a network and comprises an image acquisition device and a pressure sensing device. The image capturing device is used for shooting an image. When pressure is detected by the pressure sensing device, the image capturing device is triggered to take an image, and the poultry diagnosis device transmits the captured image to the server. The server identifies a specific part of the poultry in the captured image through an artificial intelligence image judgment model so as to judge the state of the poultry.

Description

Technical Field

[0001] The present invention relates to a technology of poultry breeding equipment, and more specifically, to an intelligent poultry auxiliary breeding system and method. Background Art

[0002] Traditional poultry farming, especially chicken farming, requires a significant amount of manual labor to perform a variety of tedious and tiring tasks. For example, farmers must manually catch and weigh chickens to ensure their stable growth and determine if adjustments to feed or environmental conditions are necessary. This labor consumes considerable time and effort and often fails to accurately reflect the health of each bird. Manually catching the birds not only causes stress, but improper stressing can also lead to death. Furthermore, farmers must constantly monitor the flock's activities, such as cleaning the coop and adjusting the living environment, to ensure adequate activity for healthy growth. These tasks require constant manual supervision, often preventing farmers from leaving the farm and focusing on other tasks, impacting work efficiency and quality of life.

[0003] On the other hand, monitoring the health of chickens, especially preventing disease in the flock, is also a challenging task. Farmers need to constantly observe the chickens' behavior to identify any signs of illness. However, this manual observation method often relies on the farmer's experience and makes it difficult to conduct comprehensive and accurate health monitoring of each chicken. Furthermore, farmers must conduct nighttime patrols to ensure the safety of the flock. This series of tasks significantly increases the labor intensity of chicken farming and can easily lead to excessive fatigue, which in turn affects work efficiency and the quality of the chickens.

[0004] Furthermore, existing chicken management relies heavily on manual operations and monitoring, and lacks automated, precise control. Consequently, chicken farmers often experience unrelieved pressure during the chicken farming process. Summary of the Invention

[0005] An object of a preferred embodiment of the present invention is to provide an intelligent poultry assisted feeding system and method, which uses remote imaging technology to determine the sex, weight, health status, etc. of poultry.

[0006] An object of a preferred embodiment of the present invention is to provide an intelligent poultry auxiliary feeding system and method for analyzing the health status of poultry based on remote imaging of their excrement.

[0007] An object of a preferred embodiment of the present invention is to provide an intelligent poultry auxiliary feeding system and method for increasing the activity of poultry.

[0008] An object of a preferred embodiment of the present invention is to provide an intelligent poultry auxiliary feeding system and method for remotely determining the average weight of poultry so as to control the weight within the electric slaughtering specification and avoid losses caused by electric slaughtering.

[0009] In view of this, a preferred embodiment of the present invention provides an intelligent poultry assisted breeding method, which includes: providing a first artificial intelligence image judgment model; collecting multiple first poultry data; generating a training set data, a verification set data, and a test set data based on the multiple first poultry data; step A: training the artificial intelligence image judgment model through the training set data and the verification set data to generate a second artificial intelligence image judgment model; step B: using the test set data to input the second artificial intelligence image judgment model to generate multiple output results; step C: when an evaluation operation result of the multiple output results is lower than a threshold value, collecting multiple second poultry data to generate the training set data, the verification set data, and the test set data, and returning to step A until the evaluation operation result reaches the threshold value; using the second artificial intelligence image judgment model that reaches the threshold value as a third artificial intelligence image judgment model; capturing an image of a poultry farm through an image input device; and identifying a state of the poultry in the image through the third artificial intelligence image judgment model.

[0010] Another preferred embodiment of the present invention provides an intelligent poultry auxiliary feeding system, which includes a server and a poultry diagnostic device. The server includes a poultry artificial intelligence image judgment model. The poultry diagnostic device is used to connect to the server via a network, wherein the poultry diagnostic device includes an image capture device and a pressure sensing device. The image capture device is used to perform image photography. When the pressure sensing device detects pressure, the image capture device is triggered to perform image photography, and the poultry diagnostic device transmits the captured image to the server. The server uses the poultry image recognition and diagnosis artificial intelligence model to identify a specific part of the poultry in the captured image, thereby determining the status of the poultry.

[0011] A preferred embodiment of the present invention proposes an intelligent poultry auxiliary feeding system, which includes a server and a poultry diagnostic device. The server includes a poultry image recognition and diagnosis artificial intelligence model. The steps for generating this poultry image recognition and diagnosis artificial intelligence model include: providing a first artificial intelligence image judgment model; collecting a plurality of first poultry data; generating a training set data, a validation set data, and a test set data based on the plurality of first poultry data; step A: training the artificial intelligence image judgment model with the training set data and the validation set data to generate a second artificial intelligence image judgment model; step B: using the test set data to input the second artificial intelligence image judgment model to generate a plurality of output results; step C: when an evaluation operation result of the plurality of output results is lower than a threshold value, collecting a plurality of second poultry data to generate the training set data, the validation set data, and the test set data, and returning to step A until the evaluation operation result reaches the threshold value; the second artificial intelligence image judgment model that reaches the threshold value is used as the poultry artificial intelligence image judgment model. The poultry diagnostic device is connected to the server via a network and includes an image input device for capturing an image in a poultry farm and transmitting it to the server, wherein the server identifies the status of the poultry in the image through the poultry artificial intelligence image judgment model.

[0012] In order to make the above and other objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The accompanying drawings are provided to enable those skilled in the art to further understand the present invention and are incorporated into and constitute part of the specification of the present invention. The accompanying drawings illustrate exemplary embodiments of the present invention and are used together with the specification to explain the principles of the present invention.

[0014] Figure 1 A system block diagram of an intelligent poultry auxiliary feeding system according to a preferred embodiment of the present invention is shown.

[0015] Figure 2 A schematic diagram illustrating a poultry diagnostic device 102 of an intelligent poultry assisted feeding system according to a preferred embodiment of the present invention is shown.

[0016] Figure 3 A schematic diagram illustrating a poultry diagnostic device 102 of an intelligent poultry assisted feeding system according to a preferred embodiment of the present invention is shown.

[0017] Figure 4 A flow chart illustrating the collection of training data for the poultry image recognition and diagnosis artificial intelligence model of the intelligent poultry assisted feeding system according to a preferred embodiment of the present invention.

[0018] Figure 5 A flowchart illustrating a method for training a poultry image recognition and diagnosis artificial intelligence model for an intelligent poultry assisted feeding system according to a preferred embodiment of the present invention.

[0019] Figure 6 A flowchart illustrating a method for training a poultry image recognition and diagnosis artificial intelligence model for an intelligent poultry assisted feeding system according to a preferred embodiment of the present invention.

[0020] Figure 7 A flowchart illustrating step S506 of the method for training a poultry image recognition and diagnosis artificial intelligence model of an intelligent poultry assisted feeding system according to a preferred embodiment of the present invention is shown.

[0021] Explanation of symbols:

[0022] 101: Server

[0023] 102: Poultry diagnostic device

[0024] 103: Image capture device

[0025] 104: Pressure sensing device

[0026] 105: Laser guidance device

[0027] 201: Hook

[0028] 202: Host

[0029] S400~S403:Steps

[0030] S501~S507: Steps

[0031] S601: Step

[0032] S701~S702:Steps DETAILED DESCRIPTION

[0033] Reference will now be made in detail to exemplary embodiments of the present invention, which are illustrated in the accompanying drawings. Wherever possible, the same reference numerals will be used in the drawings and the description to refer to the same or similar parts. The exemplary embodiments are merely one way to implement the design concepts of the present invention, and the following examples are not intended to limit the present invention.

[0034] Figure 1 The system block diagram of the intelligent poultry auxiliary feeding system according to a preferred embodiment of the present invention is shown. Figure 1This intelligent poultry assisted feeding system includes a server 101 and a poultry diagnostic device 102. Server 101 is typically located remotely, and poultry diagnostic device 102 is connected to it via, for example, a wired or wireless network. In this embodiment, server 101 includes a poultry AI image recognition model. In this embodiment, this AI image recognition model is an AI model used for image recognition.

[0035] In this embodiment, the poultry diagnostic device 102 is deployed in a poultry farm. Generally, a poultry farm may deploy multiple poultry diagnostic devices 102 depending on its size. For simplicity, this embodiment uses only one poultry diagnostic device 102 as an example. This poultry diagnostic device 102 includes an image capture device 103, a pressure sensing device 104, a laser guidance device 105, and so on. In this embodiment, the poultry diagnostic device 102 is an expandable device. Therefore, the image capture device 103, pressure sensing device 104, and laser guidance device 105 are all optional features and can be added, removed, or modified according to user needs.

[0036] Image capture device 103 is typically implemented as a digital camera or a digital video camera for capturing an image. Pressure sensing device 104 is implemented, for example, as an electronic scale. To facilitate understanding of the present invention by those skilled in the art, a poultry farm is used as an example of a chicken farm, and chickens are used as examples of poultry. Those skilled in the art will appreciate that the present invention can also be applied to other poultry, such as turkeys, geese, and ducks, and is not intended to be limiting.

[0037] Figure 2 A schematic diagram showing a poultry diagnostic device 102 of an intelligent poultry auxiliary feeding system according to a preferred embodiment of the present invention is shown. Figure 2 In this embodiment, the poultry diagnostic device 102 is designed to be suspended, with a hook 201 located above for suspending the device. When a chicken jumps onto the pressure sensing device 104, or the scale pan, the host computer 202 receives the weight change signaled by the pressure sensing device 104. At this point, the host computer 202 activates the image capture device 103 to capture an image of the chicken on the scale pan. After capturing the image, the host computer 202 transmits the captured image and the weight obtained from the scale pan to the server 101 via the network. The server 101 then activates the poultry artificial intelligence image recognition model to identify the transmitted image. Generally, the server first identifies the chicken's crown to determine the number and sex of the chickens. The crown color is then used to determine the health or abnormality of the chickens. If an abnormality is detected, such as a pale crown or an abnormal shape, an immediate warning message is issued, allowing the owner to address the issue as quickly as possible.

[0038] The above embodiment uses broiler chicken farming as an example. Weight and gender are crucial for broiler chicken farming. Since broiler chickens take approximately 30 to 35 days to raise, the standard for electric slaughter requires a weight between 2.1 and 2.3 kg. Whether too heavy or too light, roosters will experience decreased accuracy and lead to losses. However, in the later stages of chicken farming, roosters weigh more than hens, and hens are more active than roosters. Therefore, in the later stages, hens are more likely to jump onto or fly onto the scale of the pressure sensing device 104 than roosters. This situation can lead to biased weight statistics, resulting in underestimation of weight, and ultimately, misjudgment of average weight, which is out of specification, and electric slaughter failure (for example, incorrect slaughtering location leading to failed bleeding).

[0039] In this embodiment, when the statistical data contains more weight data of hens than roosters, and the poultry artificial intelligence image recognition model in the server 101 determines that the poultry on the scale of the pressure sensing device 104 is a hen, the server 101 will notify the host 202 of the poultry diagnostic device 102. The host will then activate the laser guidance device 105 to emit a guidance laser to guide the hen away from the scale of the pressure sensing device 104, thereby increasing the probability of later capturing rooster data.

[0040] While the above embodiment uses a laser beam directed under the scale pan 104 to induce hens to leave the scale pan of the pressure sensing device 104, in actual applications, in addition to directing the laser beam under the scale pan, the laser guidance device 105 can also be activated to emit a guiding laser beam, directing the laser beam onto the scale pan. This can attract hens to jump onto the scale pan, increasing the number of samples collected. Directing the laser beam under the scale pan can also attract hens away from the scale pan, preventing sampling bias. Using both methods simultaneously, in addition to achieving the aforementioned effects, can also increase the flock's activity.

[0041] The weight standards and the number of days of feeding in the above embodiments vary from country to country and also vary depending on the type of poultry being fed. Therefore, the above embodiments are merely illustrative examples and the present invention is not limited thereto.

[0042] Furthermore, during the breeding process, the health of the chickens can be assessed through their excrement. Therefore, in this embodiment, to increase the chances of capturing the chicken's excrement, when the pressure sensing device 104 detects pressure, after a predetermined period of time, the host computer 202 activates the laser guidance device 105 to guide the chicken off the scale. The host computer 202 then activates the image capture device 103 to capture the image, thereby increasing the chances of capturing the excrement. This image is also transmitted via the network to the server 101, which then activates the poultry artificial intelligence image recognition model to identify the transmitted image. If the excrement is of an abnormal color, such as green, red, or watery, a warning message will be issued, notifying the breeder to address it as soon as possible.

[0043] While the above embodiment uses the laser guide device 105 as an example, it is not a required device but rather an optional feature. In another preferred embodiment, to increase the probability of capturing chicken excrement, when the pressure sensing device 104 detects pressure, the host computer 202 waits for the pressure in the pressure sensing device 104 to drop, indicating that the chicken has left the scale pan of the pressure sensing device 104. At this point, the host computer 202 activates the image capture device 103 to capture the image, thereby also increasing the probability of capturing excrement images.

[0044] Figure 3 A schematic diagram showing a poultry diagnostic device 102 of an intelligent poultry auxiliary feeding system according to a preferred embodiment of the present invention is shown. Figure 3 In this embodiment, the poultry diagnostic device 102 comprises only a host computer 202 and an image capture device 103. In this embodiment, farmers can choose whether to use the pressure sensing device 104 based on their specific breeding conditions. For example, consider chicken farming. However, this poultry farm specializes in laying hens. Laying hens are typically caged and do not roam outdoors, so the pressure sensing device 104 is not required to determine weight. In this embodiment, a manure belt is installed beneath the cages. When the belt begins operating, the host computer 202 controls the image capture device 103 to capture an image of the poultry excrement. This image of the poultry excrement is transmitted from the host computer 202 to the server 101. The server 101 then activates the poultry artificial intelligence image recognition model to identify the transmitted image. If the excrement is abnormally colored, such as green, red, or watery, a warning message is issued, notifying the breeder to address the issue promptly.

[0045] Similarly, since caged chickens are confined and laying hens do not need to be sexed, in this embodiment, the host 202 controls the image capture device 103 to periodically take images of the chickens and transmit them to the server 101. The server 101 then activates the poultry artificial intelligence image recognition model to identify the transmitted images. If the shape or color of the crown is abnormal, a warning message will be issued to notify the breeder to handle the issue as soon as possible.

[0046] While the above embodiment uses the crest as an example for sex determination, those skilled in the art will appreciate that different methods of sex determination are employed depending on the poultry species being raised. For example, the sex of Cherry Valley ducks is determined by the curl of their tail feathers. Furthermore, male Cherry Valley ducks are typically heavier than females in later stages of mating. Therefore, the present invention is not limited to the above embodiment.

[0047] In the above embodiment, although the poultry AI image recognition model is used to determine sex, comb health, fecal health, etc. in image recognition processing, those skilled in the art will recognize that poultry weight is actually closely related to appearance. Therefore, the poultry AI image recognition model can also be used to predict or estimate poultry weight from images. Therefore, the present invention is not limited to the above applications.

[0048] The above embodiment illustrates the operation of the intelligent poultry assisted feeding system. The following embodiment will illustrate the generation method of the above poultry artificial intelligence image judgment model.

[0049] Figure 4 A flowchart showing the collection of training data for the poultry image recognition and diagnosis artificial intelligence model of the intelligent poultry auxiliary feeding system according to a preferred embodiment of the present invention is shown. Figure 4 , the steps for collecting training data include:

[0050] Step S400: Start.

[0051] Step S401: The collected poultry images are fed into an image recognition model for preliminary recognition. Generally, this collection step can be performed continuously by the image capture device 103 of the poultry diagnostic device 102. This preliminary recognition identifies which chickens are in the photo and their locations within the image.

[0052] Step S402: Labeling. Based on the chickens in the image and their positions, the comb, gender, and normal and abnormal health status are labeled. In this embodiment, the labeling process can be performed using a previously trained image recognition model, thereby accelerating the labeling process.

[0053] Step S403: Put into the training database.

[0054] Step S404: End.

[0055] Figure 5 A flowchart showing a method for training a poultry image recognition and diagnosis artificial intelligence model of an intelligent poultry auxiliary feeding system according to a preferred embodiment of the present invention is provided. Figure 5 The training method of this poultry image recognition and diagnosis artificial intelligence model includes the following steps:

[0056] Step S500: Start.

[0057] Step S501: Provide a first artificial intelligence image recognition model, such as an initial artificial intelligence image recognition model.

[0058] Step S502: Generate a training set of data, a validation set of data, and a test set of data based on the first poultry data. Figure 4 The collected image data will not be used in its entirety, but a portion will be taken out, and Figure 4 Data collection will continue. The poultry data collected will generate the above-mentioned training set data, the above-mentioned validation set data, and the above-mentioned test set data. Afterwards, in order to diversify the data and improve the stability of the model, data augmentation will be performed first, such as different brightness adjustments, different segmentation adjustments, and different rotation adjustments, to generate more training images based on the collected images of the same database. In this way, through diversified training data, the model can better adapt to changes in the real world, such as different angles, sizes, colors, etc., thereby improving the stability and accuracy of the model. In this embodiment, these training images are divided into the above-mentioned training set data, the above-mentioned validation set data, and the above-mentioned test set data in a ratio of 4:1:1.

[0059] Step S503: The first artificial intelligence image judgment model is trained using the training set data and the validation set data to generate a second artificial intelligence image judgment model. In this embodiment, the first artificial intelligence image judgment model uses, for example, a convolutional neural network (CNN). Since convolution operations are easier to extract local features (such as edges, colors, textures, etc.), they are more suitable for image processing, image feature extraction, etc. However, the present invention is not limited to other implementations, such as recurrent neural networks (RNN) and long short-term memory (LSTM).

[0060] During training, the images in the training set and the validation set are each labeled with the aforementioned features. These labels provide the true answer: the category and location of the poultry in each image. These labels help the first AI image recognition model understand the correct answer, allowing it to learn. This first AI image recognition model gradually adjusts its internal parameters (such as weights) through repeated training. During the training process, the first AI image recognition model's predictions are compared with the labeled results, and adjustments are made based on the errors.

[0061] After training is complete on the training set, the process moves to the next step: performance evaluation using the validation set. This evaluation also utilizes the mean average precision (mAP50) performance metric. Images from the validation set that fall below the threshold are returned to the first AI image recognition model for repeated training until convergence is achieved, resulting in a second AI image recognition model.

[0062] Step S504: Utilize the test set data to input the second artificial intelligence image recognition model to generate a plurality of output results.

[0063] Step S505: Determine whether it is greater than the evaluation threshold value T. If it is less than the evaluation threshold value T, proceed to step S506. In this embodiment, the evaluation threshold value T uses the mean average precision at IoU = 0.5 (mAP50) as an evaluation metric in the field of object detection. It is used to measure the accuracy of the model in identifying objects in image detection tasks. Those skilled in the art will understand that the calculation method for evaluating the threshold value of the training result can be selected according to different applications. Therefore, the present invention is not limited to the mAP50 of this embodiment.

[0064] Step S506: Recollect multiple second poultry data. If the test in step S505 fails and does not reach the threshold value, it means that the training is insufficient or the amount of data is insufficient. Figure 4 Since the image capture device 103 of the poultry diagnosis device 102 is disposed in the poultry farm and continuously collects data, a plurality of second poultry data different from the previous plurality of first poultry data can be obtained.

[0065] Step S507: Generate the training set data, validation set data, and test set data from the plurality of second poultry data. After labeling and data augmentation, the training set data, validation set data, and test set data are returned to step S503 to retrain the second AI image recognition model until it passes the evaluation threshold.

[0066] Step S508: Generate a third AI image recognition model. If all tests in step S505 pass, or if the evaluation exceeds the threshold but cannot be further improved, the training has reached convergence. At this point, further training is impossible, indicating that the training is complete. The resulting third AI image recognition model is the poultry AI image recognition model described above.

[0067] From the above examples, it can be seen that the training set data is the data set used to train the AI ​​image recognition model. The AI ​​image recognition model learns from this data and adjusts its internal parameters to improve prediction accuracy. From the perspective of a typical learner, it can be likened to a textbook. The validation set data is used to evaluate the performance of the AI ​​image recognition model during the training process and to help adjust its parameters. It does not participate in the training of the AI ​​image recognition model, but is used to determine its learning progress. From the perspective of a typical learner, it can be likened to a set of exercises with solutions. The test set data is a dataset used to evaluate the performance of the AI ​​image recognition model. This data is not used at all during the training and validation of the AI ​​image recognition model and is used to evaluate the AI ​​image recognition model's performance on new data. From the perspective of a typical learner, it can be likened to an unanswered question that tests learning outcomes. Therefore, in the above examples, it can be seen that the training set data and the validation set data both have feature-labeled images, while the test set data does not.

[0068] In the above embodiment, the images in the test set that were incorrectly recognized were re-labeled and used as training data, and the original second artificial intelligence image recognition model was retrained to generate a third artificial intelligence image recognition model. Therefore, through this repeated training method, the third artificial intelligence image recognition model, namely the poultry artificial intelligence image recognition model, was further strengthened.

[0069] Figure 6 A flowchart showing a method for training a poultry image recognition and diagnosis artificial intelligence model of an intelligent poultry auxiliary feeding system according to a preferred embodiment of the present invention is provided. Figure 6The training method of the poultry image recognition and diagnosis artificial intelligence model further includes the following steps before step S503:

[0070] Step S601: Utilize a fourth AI image recognition model to perform feature pre-labeling on images within the training set and validation set. In this embodiment, the fourth AI image recognition model can be a previous version of the poultry AI image recognition model. This allows the learning results of the previous version of the poultry image recognition and diagnosis AI model to be transferred to the subsequently trained third AI image recognition model, facilitating rapid training of the AI ​​image recognition model.

[0071] Figure 7 Flowchart showing step S506 of the method for training the poultry image recognition and diagnosis artificial intelligence model of the intelligent poultry auxiliary feeding system according to a preferred embodiment of the present invention. Figure 7 Step S506 of the training method for the poultry image recognition and diagnosis artificial intelligence model further includes the following steps:

[0072] Step S701: Determine the difference between the output evaluation value and the evaluation threshold value T.

[0073] Step S702: Determine the number of second poultry data sets to collect based on the difference. For example, if the difference between the evaluation value and the evaluation threshold T is within 5%, the number of second poultry data sets can be half the number of first poultry data sets. If the difference between the evaluation value and the evaluation threshold T is between 5% and 10%, the number of second poultry data sets can be equal to the number of first poultry data sets. If the difference between the evaluation value and the evaluation threshold T is greater than 10%, the number of second poultry data sets can be double the number of first poultry data sets. This reduces the number of training cycles and the amount of training computation.

[0074] In summary, the preferred embodiment of the present invention can not only effectively improve the accuracy of automated monitoring in the poultry breeding process, but also significantly improve the efficiency of breeding management by combining the poultry image recognition and diagnosis artificial intelligence model with data enhancement technology. This method ensures the high-precision recognition ability of the model when facing different situations by classifying and processing multiple poultry data, and repeatedly training and adjusting the training set, validation set and test set data. When an erroneous result appears in the test set data, the model can use data enhancement technology to generate more training samples, thereby optimizing the performance of the model. This not only improves the accuracy of the recognition results, but also helps farms to achieve real-time monitoring of the growth status of poultry, thereby achieving multiple advantages such as refined management, reduced labor costs, and improved production efficiency. It has important application value to the poultry breeding industry.

[0075] The specific embodiments described in the detailed description of the preferred embodiments are merely for the purpose of illustrating the technical content of the present invention and are not intended to limit the present invention to these embodiments. Any modifications and variations made without departing from the spirit of the present invention and the following claims are intended to fall within the scope of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

Claims

1. An intelligent poultry assisted feeding method, comprising: Providing a first artificial intelligence image judgment model; Collect multiple first poultry data; Generating a training set of data, a validation set of data, and a test set of data according to the plurality of first poultry data; Step A: Training the first artificial intelligence image recognition model using the training set data and the validation set data to generate a second artificial intelligence image recognition model; Step B: Using the test set data to input the second artificial intelligence image recognition model to generate multiple output results; Step C: When an evaluation result of the plurality of output results is lower than a threshold value, collecting a plurality of second poultry data to generate the training set data, the validation set data, and the test set data, and returning to Step A until the evaluation result reaches the threshold value; using the second artificial intelligence image recognition model that reaches the threshold value as a third artificial intelligence image recognition model; capturing an image of a poultry farm through an image input device, wherein the poultry farm contains a plurality of poultry; and The third artificial intelligence image judgment model is used to identify the status of one of the multiple poultry in the image.

2. The intelligent poultry assisted feeding method according to claim 1, wherein: Training the artificial intelligence image recognition model using the training set data and the validation set data to generate a second artificial intelligence image recognition model includes at least one of the following steps: Performing different brightness adjustments on the images of the training set data and the validation set data to increase the amount of the training set data and the validation set data; Performing different cut adjustments on the images of the training set data and the validation set data to increase the amount of the training set data and the validation set data; and The images of the training set data and the validation set data are rotated differently to increase the amount of the training set data and the validation set data.

3. The intelligent poultry assisted feeding method according to claim 1, wherein: The image input device is used to capture the image in the poultry farm, including: When a pressure sensing device senses pressure, an image is captured to obtain the image.

4. The intelligent poultry assisted feeding method according to claim 1, wherein: Identifying the state of the poultry in the image using the third artificial intelligence image judgment model includes: Determine the sex of poultry by looking at their crown.

5. The intelligent poultry assisted feeding method according to claim 1, wherein: Identifying the state of the poultry in the image using the third artificial intelligence image judgment model includes: The health status of poultry can be judged by the crown of the poultry.

6. The intelligent poultry assisted feeding method according to claim 1, wherein: When the image includes multiple poultry, it also includes: identifying a head of each of the plurality of poultry to obtain a head pattern of each of the plurality of poultry; The crown of each of the plurality of poultry head patterns is taken out and compared with the image to find the corresponding poultry.

7. The intelligent poultry assisted feeding method according to claim 1, wherein: Capturing the image in the poultry farm through the image input device further includes: When a pressure sensing device senses pressure, an image is taken after the pressure of the pressure sensing device drops, thereby increasing the probability of obtaining an image of poultry excrement.

8. The intelligent poultry assisted feeding method according to claim 7, wherein: When the pressure sensing device senses pressure, the imaging is performed after the pressure of the pressure sensing device drops, thereby increasing the probability of capturing an image of poultry excrement. The method further includes: When the poultry has been on the pressure sensing device for a preset time, a laser guiding device is started to emit a laser to guide the poultry on the pressure sensing device to leave the pressure sensing device.

9. The intelligent poultry assisted feeding method according to claim 1, wherein: Capturing the image in the poultry farm through the image input device further includes: provide a manure belt within the poultry farm; and When the excrement dragging belt starts to run, a video shooting is performed to obtain an image of poultry excrement.

10. The intelligent poultry assisted feeding method according to claim 1, wherein: Capturing the image in the poultry farm through the image input device further includes: When a pressure sensing device senses weight, an image is captured, and the sex of the poultry in the image is identified using the third artificial intelligence image recognition model, and the sex and corresponding weight are recorded; and When the number of recorded poultry of a first sex is greater than the number of recorded poultry of a second sex, and the sex of the poultry on the pressure sensing device is identified as the first sex, the poultry of the first sex is guided out of the pressure sensing device by a laser guiding device to increase the probability of sampling the poultry of the second sex.

11. The intelligent poultry assisted feeding method according to claim 1, wherein: The images in the training set data and the images in the validation set data have feature annotations.

12. The intelligent poultry assisted feeding method according to claim 11, further comprising: A fourth artificial intelligence image judgment model is used to perform feature annotation on the images within the training set data and the images within the verification set data.

13. An intelligent poultry auxiliary feeding system, comprising: a server comprising a poultry artificial intelligence image judgment model; as well as A poultry diagnostic device, configured to connect to the server via a network, wherein the poultry diagnostic device comprises: an image capturing device for taking an image; and a pressure sensing device, When the pressure sensing device detects pressure, the image capturing device is triggered to take the image, and the poultry diagnostic device transmits the captured image to the server. The server uses the poultry artificial intelligence image judgment model to identify a specific part of the poultry in the captured image to determine the status of the poultry.

14. The intelligent poultry auxiliary feeding system according to claim 13, wherein: The poultry diagnostic device is further used to: The pressure data and the image captured by the pressure sensing device are transmitted to the server, The server records poultry data according to the pressure value and the number and sex of the poultry identified by the image, and determines the growth status according to the statistical results of the plurality of poultry data.

15. The intelligent poultry auxiliary feeding system according to claim 14, wherein: The poultry diagnostic device further comprises: A laser guiding device for emitting a laser, wherein: When the corresponding data of a first sex of the recorded poultry is greater than the corresponding data of a second sex, and When the poultry in the image is determined to be of the first gender, the laser guiding device is activated to guide the poultry on the pressure sensing device away from the pressure sensing device.

16. The intelligent poultry auxiliary feeding system according to claim 13, wherein: The poultry diagnostic device further comprises: a laser guiding device for emitting a laser, wherein the poultry diagnostic device is further configured to: When the pressure sensing device detects pressure, the laser guiding device is activated to guide the poultry on the pressure sensing device away from the pressure sensing device, and the image capturing device is activated to capture an image of the poultry's excrement and transmit it to the server. Among them, the poultry artificial intelligence image judgment model determines the health status of poultry based on the excrement image.

17. The intelligent poultry auxiliary feeding system according to claim 13, wherein: When the pressure sensing device senses pressure, the image is photographed after the pressure of the pressure sensing device drops, thereby increasing the probability of capturing the image of poultry excrement.

18. The intelligent poultry auxiliary feeding system according to claim 13, wherein: The training steps for the poultry AI image recognition model include: Providing a first artificial intelligence image judgment model; Generating a training set of data, a validation set of data, and a test set of data according to the plurality of first poultry data; Step A: Training the first artificial intelligence image recognition model using the training set data and the validation set data to generate a second artificial intelligence image recognition model; Step B: Using the test set data to input the second artificial intelligence image recognition model to generate multiple output results; Step C: When an evaluation result of the plurality of output results is lower than a threshold value, collecting a plurality of second poultry data to generate the training set data, the validation set data, and the test set data, and returning to Step A until the evaluation result reaches the threshold value; The second artificial intelligence image judgment model that reaches the threshold value is used as the poultry artificial intelligence image judgment model.

19. The intelligent poultry auxiliary feeding system according to claim 18, wherein: Training the artificial intelligence image recognition model using the training set data and the validation set data to generate a second artificial intelligence image recognition model includes at least one of the following steps: Performing different brightness adjustments on the images of the training set data and the validation set data to increase the amount of the training set data and the validation set data; Performing different cut adjustments on the images of the training set data and the validation set data to increase the amount of the training set data and the validation set data; and The images of the training set data and the validation set data are rotated differently to increase the amount of the training set data and the validation set data.

20. The intelligent poultry assisted feeding system according to claim 19, further comprising: A fourth artificial intelligence image judgment model is used to feature-label the images within the training set data and the images within the verification set data to train the poultry artificial intelligence image judgment model.

21. An intelligent poultry auxiliary feeding system, comprising: A server includes an artificial intelligence model for poultry image recognition and diagnosis, wherein the steps of generating the poultry artificial intelligence image recognition model include: Providing a first artificial intelligence image judgment model; Generating a training set of data, a validation set of data, and a test set of data according to the plurality of first poultry data; Step A: Training the first artificial intelligence image recognition model using the training set data and the validation set data to generate a second artificial intelligence image recognition model; Step B: Using the test set data to input the second artificial intelligence image recognition model to generate multiple output results; Step C: When an evaluation result of the plurality of output results is lower than a threshold value, collecting a plurality of second poultry data to generate the training set data, the validation set data, and the test set data, and returning to Step A until the evaluation result reaches the threshold value; using the second artificial intelligence image recognition model that reaches the threshold value as the poultry artificial intelligence image recognition model; and A poultry diagnostic device is connected to a server via a network, comprising an image input device for capturing an image within a poultry farm and transmitting the image to the server, wherein the server identifies a state of the poultry in the image through the poultry image recognition and diagnosis artificial intelligence model.

22. The intelligent poultry assisted feeding system according to claim 21, wherein: Training the artificial intelligence image recognition model using the training set data and the validation set data to generate a second artificial intelligence image recognition model includes at least one of the following steps: Performing different brightness adjustments on the images of the training set data and the validation set data to increase the amount of the training set data and the validation set data; Performing different cut adjustments on the images of the training set data and the validation set data to increase the amount of the training set data and the validation set data; and The images of the training set data and the validation set data are rotated differently to increase the amount of the training set data and the validation set data.

23. The intelligent poultry assisted feeding system according to claim 21, wherein: The steps for generating the poultry artificial intelligence image recognition model further include: A fourth artificial intelligence image judgment model is used to feature-label the images within the training set data and the images within the verification set data to train the poultry artificial intelligence image judgment model.

24. The intelligent poultry assisted feeding system according to claim 21, wherein: The poultry diagnostic device further comprises: A pressure sensing device, wherein when the pressure sensing device detects pressure, the image capture device is triggered to take the image. The poultry diagnostic device transmits the pressure data and the image captured by the pressure sensing device to the server. The server records poultry data according to the pressure value and the number and sex of the poultry identified by the image, and determines the growth status according to the statistical results of the plurality of poultry data.

25. The intelligent poultry assisted feeding system according to claim 21, wherein: The poultry diagnostic device further comprises: A laser guiding device for emitting a laser, wherein: When the corresponding data of a first gender of the recorded poultry is greater than the corresponding data of a second gender, and the poultry in the image is determined to be of the first gender The laser guide device is activated to guide the poultry on the pressure sensing device away from the pressure sensing device.

26. The intelligent poultry assisted feeding system according to claim 21, wherein: The poultry diagnostic device further comprises: A laser guide device for emitting a laser, wherein the poultry diagnostic device is further used to When the pressure sensing device detects pressure, the laser guiding device is activated to guide the poultry on the pressure sensing device away from the pressure sensing device, and the image capturing device is activated to capture an image of the poultry's excrement and transmit it to the server. Among them, the poultry artificial intelligence image judgment model determines the health status of poultry based on the excrement image.

27. The intelligent poultry assisted feeding system according to claim 21, wherein: When the pressure sensing device senses pressure, the image is photographed after the pressure of the pressure sensing device drops, thereby increasing the probability of capturing the image of poultry excrement.

Images