Poultry disease detection system based on knowledge graph

Abstract

The invention relates to a poultry disease detection system based on a knowledge graph. The poultry disease detection system mainly comprises a data acquisition module and a disease recognition module, wherein the data acquisition module is used for reading poultry dissection pictures and corresponding text descriptions; and the disease recognition module is used for recognizing symptom features by adopting a Clip model according to the poultry dissection pictures and the corresponding text description, and acquiring corresponding disease names, treatment methods and environmental influence factors based on the symptom features and a knowledge graph. According to the application, through cooperative operation of the image recognition model and the knowledge graph, rapid, accurate and comprehensive detection of the poultry respiratory diseases can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of poultry disease detection, and more particularly to a poultry disease detection system based on a knowledge graph. Background Art

[0002] Poultry respiratory diseases are a common epidemic disease in intensive farming and pose serious risks. Timely and accurate detection of such diseases is crucial to the healthy development of poultry farming.

[0003] However, due to the complex and diverse clinical manifestations of poultry respiratory diseases and their significant overlap, the traditional method of relying on visual observation of poultry phenotypic information or autopsy information requires high professional knowledge and rich experience. It is difficult for non-professionals to make accurate judgments, and it is easy to make misdiagnoses or miss diagnosis.

[0004] Secondly, pathological testing has a significant lag effect. That is, when chickens are suspected of being sick, they need to be sent to a veterinarian for diagnosis. Since the process of sending and autopsy sick chickens is time-consuming, it lacks timeliness. During this period, the disease may spread rapidly among poultry, leading to larger-scale infection.

[0005] In addition, the existing methods have weak multi-source data integration capabilities, and there is an island effect for multimodal data such as symptom data, environmental parameters (ammonia concentration, temperature and humidity), gene sequences, and medication records, and lack the ability to model knowledge based on spatiotemporal correlations. Summary of the Invention

[0006] In view of this, in order to at least partially solve the above technical problems, the present invention provides a poultry disease detection system based on knowledge graph, which can significantly improve the accuracy and real-time performance of poultry disease diagnosis.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] A poultry disease detection system based on knowledge graph, including:

[0009] A data acquisition module for reading poultry autopsy images and corresponding text descriptions;

[0010] The disease recognition module is used to identify symptom characteristics using the Clip model based on poultry autopsy images and corresponding text descriptions, and to obtain the corresponding disease name, treatment method and environmental influencing factors based on the knowledge graph based on the symptom characteristics.

[0011] In an optional embodiment, the Clip model introduces a cross-attention mechanism.

[0012] In an optional embodiment, the knowledge graph is constructed by the following steps:

[0013] A joint entity relationship extraction model for poultry disease diagnosis and treatment text based on BERT and dual-pointer annotation automatically extracts four entities: poultry disease, autopsy symptoms, environmental factors, and treatment methods from text data such as poultry disease-related literature, books, and actual case studies.

[0014] According to the knowledge structure of disease diagnosis and treatment, attribute labels are divided into different categories for triple entities and relationships;

[0015] Import entities, relationships, and attribute labels into the Neo4j graph database to generate a knowledge graph.

[0016] In an optional embodiment, when training Clip, entity text in the knowledge graph is used for annotation.

[0017] In an optional embodiment, a user login module is further included to authenticate the user.

[0018] In an optional embodiment, a result display module is further included for displaying the disease name, treatment method and environmental influencing factors obtained based on the knowledge graph.

[0019] The present invention discloses a poultry disease detection system based on a knowledge graph. After the user uploads a poultry autopsy image, the system can respond immediately, analyze quickly, and provide key information such as disease diagnosis results and treatment methods to the user in a very short time.

[0020] Compared with the existing technology, the present invention integrates the image detection model and the knowledge graph, and through efficient data transmission and collaborative work, realizes the optimization of the complete process from uploading the image to the accurate presentation of disease diagnosis results and related treatment information. At the same time, it ensures that the system functions are simple and easy to use, and the user operation is standardized and convenient, thereby improving the overall detection service quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0022] Figure 1 This is a schematic diagram of the structure of the poultry disease detection system based on the knowledge graph of the present invention;

[0023] Figure 2 This is a flowchart of the use of the poultry disease detection system based on the knowledge graph of the present invention. DETAILED DESCRIPTION

[0024] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0025] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0026] The embodiment of the present invention discloses a poultry disease detection system based on a knowledge graph, which can be used to detect respiratory diseases in chickens and the like. The system includes:

[0027] A data acquisition module for reading poultry autopsy images and corresponding text descriptions;

[0028] Traditional detection methods often rely on single-modality data. When using text-based diagnosis, it is impossible to obtain autopsy images in real time, making it difficult to timely understand the pathological characteristics of the chickens and quickly determine the diseases they suffer from. When relying solely on autopsy images, there is a lack of textual information support. For example, it is difficult to accurately match the characteristics of lung lesions with textual information such as treatment methods, resulting in large errors in knowledge graph diagnosis.

[0029] This application considers both poultry autopsy images and corresponding text descriptions, which can meet the prevention and control needs of accurate and efficient detection in poultry farming.

[0030] The disease recognition module is used to identify symptom characteristics using the Clip model based on poultry autopsy images and corresponding text descriptions, and to obtain the corresponding disease name, treatment method and environmental influencing factors based on the knowledge graph based on the symptom characteristics.

[0031] This application first outputs lesion information based on chicken autopsy images, then uses knowledge graphs to infer the chicken respiratory disease category and provide corresponding diagnosis and treatment plans, thereby achieving accurate detection of chicken respiratory diseases. This invention can be widely used in poultry farms. Breeding companies and farmers can use this system to diagnose respiratory diseases in poultry, thereby effectively reducing the risk of group infection and ensuring breeding yield and quality.

[0032] In one embodiment, the knowledge graph is constructed by the following steps:

[0033] JEER_PD, a joint entity relationship extraction model for poultry disease diagnosis and treatment text based on BERT and dual-pointer annotation, automatically extracts four entities: poultry diseases, autopsy symptoms, environmental factors, and treatment methods from text data such as poultry disease-related literature, books, and actual case studies. The JEER_PD model enhances the completeness of entities extracted related to poultry respiratory diseases and effectively reduces the extraction of irrelevant information.

[0034] In-depth analysis of the knowledge structure of disease diagnosis and treatment, and classification of attribute labels for triple entities and relationships according to different categories;

[0035] The Neo4j graph database is used to store knowledge. The entities, relationships, and attribute labels obtained by the text extraction model are imported into the Neo4j graph database to construct a knowledge graph of poultry respiratory diseases.

[0036] This knowledge graph is constructed through data integration and actual diagnosis of cases, using a text extraction model to build a more comprehensive and professional poultry respiratory disease knowledge graph database, which is the core support for the disease diagnosis data of the present invention.

[0037] In an optional embodiment, a pre-trained multimodal fusion detection algorithm CLIP is used to construct a classification model for poultry respiratory disease autopsy images; the core principle of the CLIP model is that it can process image and text information at the same time, and achieve the image classification task by associating image features with text features.

[0038] When training Clip, we first obtained poultry disease image data from literature, professional books, case studies, and actual farms. Through actual collection and data collection, we integrated a large number of autopsy images used to effectively identify chicken respiratory diseases, effectively ensuring the professionalism and comprehensiveness of the data source.

[0039] The poultry respiratory disease autopsy images are then preprocessed using technologies such as Mixup and Mosaic, and the sample images are annotated with entity text from the knowledge graph. The clip model learns from a large number of poultry respiratory disease autopsy images and their text descriptions, and then outputs the corresponding text entities based on the image detection, which is used for system search and reasoning.

[0040] The core of the CG-CLIP model in this application lies in the fusion of a CBAM-enhanced visual encoder and a GPT-2-powered text encoder to effectively improve cross-modal retrieval accuracy. Specifically, the CBAM-ResNet50 architecture is used as the image encoder, leveraging a channel-spatial dual-domain attention mechanism to enhance feature extraction from pathological images. Simultaneously, a lightweight GPT-2 is used as the text encoder, leveraging its autoregressive properties to capture the long-range semantics of medical text and enhance the semantic parsing of medical terminology.

[0041] By optimizing the dual encoder through contrastive learning, the model achieves fine-grained alignment of image and text features in a shared embedding space. This model has significant advantages and can effectively improve the sensitivity of histopathological features.

[0042] In a specific embodiment, a poultry disease detection system based on a knowledge graph is constructed, and the structure is as follows: Figure 1 As shown, the construction steps include:

[0043] 1. Development environment and module selection;

[0044] The detection platform and model building were developed using Python 3.8. Development was completed using Python's relevant functional modules, with PyQt used for the graphical user interface. The detection system design consists of four modules: user login, data reading, model recognition, and result display.

[0045] 2. User login module;

[0046] This module is responsible for verifying user identities, ensuring that only authorized users can use the testing platform and protecting data security and privacy. Users need to enter their pre-registered username and password, and only after the system verifies that they have passed can they access the subsequent operation interface.

[0047] 3. Data acquisition module:

[0048] Users upload poultry autopsy images to the platform. The data reading module receives and interprets the image data, passing it to the image detection model for analysis and processing. The module also reads relevant textual information entered by the user, such as information about the chicken's breeding environment, to be considered in knowledge graph queries.

[0049] 4.Disease identification module;

[0050] The image detection model Clip recognizes uploaded images and extracts symptom information, such as lung lesions and tracheal abnormalities, from the images. These symptoms are then sent to the knowledge graph. Searches within the knowledge graph, combining entity relationships and attribute labels within the knowledge graph, reveal disease information corresponding to the symptoms.

[0051] 5. Result display module;

[0052] The platform will output the disease name obtained from the query and display it to the user. It can also display information such as treatment methods, preventive measures, and possible environmental triggers related to the disease, providing users with comprehensive chicken respiratory disease detection results and prevention and control recommendations.

[0053] This application integrates the image detection model and the knowledge graph. Through efficient data transmission and collaborative work, it optimizes the entire process from image uploading to accurate presentation of disease diagnosis results and related treatment information. At the same time, it ensures that the system functions are simple and easy to use, and the user operations are standardized and convenient, thereby improving the overall detection service quality.

[0054] For example, in a chicken farm, when a sick chicken is found, a photo is taken and uploaded, and the system instantly starts the test and gives the result quickly, thus buying precious time for timely disease prevention and control. Figure 2 ; It includes: logging into the system; selecting "Upload together"; selecting sick chickens; opening the skin on the neck, separating the muscle tissue, and exposing the trachea; dissecting downwards along the trachea, exposing the bronchi, liver, air sacs, nasal cavity, larynx and other respiratory organs in turn; taking photos; uploading photos; system processing; viewing the results.

[0055] The image detection model in this application identifies lesion features and transmits the data to the knowledge graph. Through efficient query and reasoning, the knowledge graph quickly and accurately finds corresponding disease and treatment information. For example, if a specific lung lesion is detected, the knowledge graph can quickly match the disease and provide an appropriate treatment plan. The two work together to significantly improve accuracy, achieving precise alignment of symptoms and autopsy images.

[0056] Advantages of this application include:

[0057] 1. Improve detection timeliness: It can quickly detect respiratory diseases in chickens, greatly shortening the detection cycle. The process from uploading images to outputting disease diagnosis results can be completed within minutes, which is much more efficient than traditional veterinary detection methods.

[0058] 2. Improve detection accuracy: By building a precise knowledge graph and aligning images and text, we can more accurately identify respiratory diseases in chickens and accurately distinguish the subtle differences between similar diseases.

[0059] 3. Ensure data reliability: The data used are all derived from authoritative literature, professional books and actual farm data authenticated by professional veterinarians, ensuring the comprehensiveness and reliability of the data and providing stronger support for testing.

[0060] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method description.

[0061] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A poultry disease detection system based on knowledge graph, characterized in that: include: A data acquisition module for reading poultry autopsy images and corresponding text descriptions; The disease recognition module is used to identify symptom characteristics using the Clip model based on poultry autopsy images and corresponding text descriptions, and to obtain the corresponding disease name, treatment method and environmental influencing factors based on the knowledge graph based on the symptom characteristics.

2. The poultry disease detection system according to claim 1, characterized in that: The Clip model introduces a cross-attention mechanism.

3. The poultry disease detection system according to claim 1, characterized in that: The knowledge graph is constructed through the following steps: A joint entity relationship extraction model for poultry disease diagnosis and treatment text based on BERT and dual-pointer annotation automatically extracts four entities: poultry disease, autopsy symptoms, environmental factors, and treatment methods from text data such as poultry disease-related literature, books, and actual case studies. According to the knowledge structure of disease diagnosis and treatment, attribute labels are divided into different categories for triple entities and relationships; Import entities, relationships, and attribute labels into the Neo4j graph database to generate a knowledge graph.

4. The poultry disease detection system according to claim 3, characterized in that: When training Clip, it is annotated using entity text in the knowledge graph.

5. The poultry disease detection system according to claim 1, characterized in that: A user login module is also included to authenticate users.

6. The poultry disease detection system according to claim 1, characterized in that: It also includes a result display module for displaying the disease names, treatment methods and environmental influencing factors obtained based on the knowledge graph.

Images