Search

CN-121982745-A - Intelligent identification terminal of self-adaptation thing networking

CN121982745ACN 121982745 ACN121982745 ACN 121982745ACN-121982745-A

Abstract

The invention belongs to the technical field of intelligent recognition and detection, and discloses an intelligent recognition terminal of a self-adaptive Internet of things, which comprises a shell, a data acquisition module, an edge calculation module, an information fusion and decision module and an Internet of things communication module, wherein the data acquisition module is used for synchronously acquiring behavior inertial data and visual image data of a target animal, the edge calculation module is in communication connection with the data acquisition module and is used for running a trained behavior recognition model and a visual representation recognition model and carrying out real-time processing and recognition on the acquired data, the information fusion and decision module is integrated with the edge calculation module, and the Internet of things communication module is used for uploading the state judgment and terminal data to a remote management platform. According to the scheme, inertial motion and visual image data are synchronously acquired, and real-time processing and information fusion are carried out on the terminal by utilizing edge calculation and an AI model locally, so that the accuracy, the instantaneity and the environmental adaptability of animal state identification are effectively improved.

Inventors

  • HU SIPENG

Assignees

  • 深圳市同晟基业科技有限公司

Dates

Publication Date
20260505
Application Date
20260122

Claims (10)

  1. 1. An intelligent identification terminal of a self-adaptive Internet of things is characterized by comprising: a housing (100); The data acquisition module (200) is used for synchronously acquiring behavior inertial data and visual image data of the target animal; The edge computing module (300) is in communication connection with the data acquisition module (200) and is used for running the trained behavior recognition model and the visual representation recognition model and carrying out real-time processing and recognition on the acquired data; The information fusion and decision module (400) is integrated with the edge calculation module (300) and is used for carrying out fusion analysis on the behavior recognition result and the visual representation recognition result according to preset rules to generate a state judgment conclusion of the target animal; and the internet of things communication module (500) is used for uploading the state judgment and terminal data to a remote management platform.
  2. 2. The intelligent recognition terminal of the self-adaptive Internet of things according to claim 1, wherein the data acquisition module (200) comprises: the inertial measurement unit (210) is used for acquiring triaxial acceleration, triaxial angular velocity and triaxial magnetic field data of the body surface of the target animal; An image acquisition unit (220) for acquiring image data of a specific physiological site of a target animal; Wherein the inertial measurement unit (210) and the image acquisition unit (220) are data synchronized based on a uniform time stamp.
  3. 3. The intelligent recognition terminal of the self-adaptive Internet of things according to claim 2, wherein the inertial measurement unit (210) is integrated in a wearable wearing structure; the image acquisition unit (220) comprises a camera module and an adjustable bracket for fixing the module.
  4. 4. The intelligent recognition terminal of the self-adaptive Internet of things according to claim 1, wherein the behavior recognition model is a time sequence classification model based on a long-short-term memory network and is used for classifying a behavior inertia data sequence into a plurality of preset behavior categories; the visual representation recognition model is an image classification model based on a YOLO architecture and is used for carrying out feature extraction and state classification on visual image data.
  5. 5. The intelligent recognition terminal of the self-adaptive Internet of things according to claim 1, wherein the preset rule of the information fusion and decision module (400) is a fusion judgment method based on a weighted score or a decision mapping table.
  6. 6. The intelligent recognition terminal of the self-adaptive Internet of things according to claim 1, comprising a man-machine interaction module (600), wherein the man-machine interaction module (600) comprises a display screen (610) and an audible-visual annunciator (620) for locally displaying state judgment and system information.
  7. 7. The intelligent recognition terminal of the self-adaptive Internet of things according to claim 1, wherein the intelligent recognition terminal comprises a power management module (700), and the power management module (700) comprises a rechargeable battery and a power management circuit and is used for providing continuous power supply for the terminal.
  8. 8. The intelligent recognition terminal of the adaptive Internet of things according to claim 1, wherein the intelligent recognition terminal comprises a data storage unit (800) for caching or persistently storing the collected original data, recognition results and system logs.
  9. 9. The intelligent recognition terminal of the self-adaptive Internet of things according to claim 1, wherein the edge computing module (300) is an embedded artificial intelligent computing platform.
  10. 10. The intelligent recognition terminal of the self-adaptive Internet of things according to claim 1, wherein the inertial measurement unit (210) is a wireless inertial measurement sensor; the image acquisition unit (220) is a high-definition camera module.

Description

Intelligent identification terminal of self-adaptation thing networking Technical Field The invention relates to the technical field of intelligent recognition and detection, in particular to an intelligent recognition terminal of a self-adaptive Internet of things. Background In the fields of animal breeding management, health monitoring, breeding regulation and control and the like, accurately and timely identifying the key physiological states (such as oestrus states) of animals has important significance for improving production benefits and animal welfare. At present, common monitoring methods mainly depend on manual observation, single sensor threshold judgment or a visual identification system based on cloud processing, and have obvious defects in practical application, namely firstly, the method depends on low manual observation efficiency, is high in subjectivity and is difficult to realize all-weather continuous monitoring, secondly, the method only based on threshold value of motion inertia data (such as acceleration) can detect obvious behavior change, but cannot distinguish different physiological states in similar motion modes, is easy to interfere and is single in characteristic dimension, thirdly, the method only depends on visual image identification is greatly influenced by factors such as illumination conditions, shielding and shooting angles, and the like, and is poor in stability in complex environments, and fourthly, the traditional scheme adopts an acquisition-uploading-cloud analysis mode, is limited by network bandwidth, transmission delay and cloud service availability, is difficult to meet scene requirements with high real-time requirements, and brings data security and privacy risks. In addition, the mode often fails to effectively fuse the multi-mode information, so that the state judgment accuracy is limited, the false alarm rate is high, and a reliable basis cannot be provided for accurate decision. Therefore, there is a need for an adaptive internet of things recognition system capable of locally completing multi-mode data synchronous acquisition, real-time intelligent analysis and fusion judgment at a terminal, so as to solve the above-mentioned prominent pain points in terms of real-time performance, accuracy, reliability and environmental adaptability. Disclosure of Invention The invention aims to provide an intelligent identification terminal of a self-adaptive Internet of things so as to solve the problems in the background technology. In order to achieve the above purpose, the present invention provides the following technical solutions: an intelligent identification terminal of self-adaptation thing networking includes: A housing; the data acquisition module is used for synchronously acquiring behavior inertial data and visual image data of the target animal; The edge computing module is in communication connection with the data acquisition module and is used for running the trained behavior recognition model and the visual representation recognition model and carrying out real-time processing and recognition on the acquired data; The information fusion and decision module is integrated with the edge calculation module and is used for carrying out fusion analysis on the behavior recognition result and the visual characterization recognition result according to a preset rule to generate a state judgment conclusion of the target animal; And the communication module of the Internet of things is used for uploading the state judgment and terminal data to a remote management platform. Preferably, the data acquisition module comprises: the inertial measurement unit is used for acquiring triaxial acceleration, triaxial angular velocity and triaxial magnetic field data of the body surface of the target animal; the image acquisition unit is used for acquiring image data of a specific physiological part of the target animal; the inertial measurement unit and the image acquisition unit perform data synchronization based on uniform time stamps. Preferably, the inertial measurement unit is integrated in a wearable wearing structure; the image acquisition unit comprises a camera module and an adjustable bracket for fixing the module. Preferably, the behavior recognition model is a time sequence classification model based on a long-term and short-term memory network, and is used for classifying the behavior inertia data sequence into a plurality of preset behavior categories; the visual representation recognition model is an image classification model based on a YOLO architecture and is used for carrying out feature extraction and state classification on visual image data. Preferably, the preset rule of the information fusion and decision module is a fusion decision method based on a weighted score or a decision mapping table. Preferably, the system comprises a man-machine interaction module, wherein the man-machine interaction module comprises a display screen and an audible-visual a