AI Pose Estimation for Companion Animal Joint Abnormality Diagnosis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
There is a need for an efficient and accurate method to predict and diagnose joint conditions in companion animals, particularly patellar luxation, to prevent the disease through early detection and management, as existing methods often rely on incorrect information and face challenges with long waiting times and inadequate service at veterinary hospitals.
Innovation Solution
A service providing system using a convolution neural network (CNN) to analyze frame-by-frame image data from a video, extracting predicted joint data, comparing it to standard data, and generating diagnosis results to determine joint abnormalities, which includes separating the subject from the background, setting bounding boxes, and analyzing variables like step height, leg angle, and movement differences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If veterinary hospitals are used for diagnosis, then accurate joint condition assessment can be obtained, but long waiting times and heavy workload reduce service quality and accessibility
Solution Approach 1:
The patent creates a virtual copy of the veterinary hospital's diagnostic capability through an AI-based pose estimation system. The system replicates the joint condition assessment function by analyzing video data of animal movement, generating diagnostic results that mirror what would be obtained at a physical veterinary hospital, thereby eliminating waiting times while maintaining diagnostic accuracy.
Solution Approach 2:
The patent introduces an AI-based pose estimation system as an intermediary between the user and the veterinary hospital. This intermediary processes video data and provides preliminary joint condition assessments, reducing the need for direct hospital visits and minimizing waiting times while preserving access to accurate diagnostic capabilities.
2Ease of operation
If information from nearby people or Internet is used, then immediate advice can be obtained, but incorrect information leads to improper treatment
Solution Approach 1:
The patent enables users to perform self-diagnosis of their animals' joint conditions by uploading video data to the AI system. The system automatically analyzes the movement patterns and provides diagnostic results, allowing users to obtain reliable information independently without relying on potentially incorrect advice from others, while maintaining the convenience of home-based assessment.
Solution Approach 2:
The patent replaces the mechanical system of human-to-human information transmission (veterinarians, pet owners, Internet sources) with an AI-based automated analysis system. This substitution eliminates the propagation of incorrect information while maintaining ease of access, as the AI system provides consistent, evidence-based diagnostic results based on objective movement analysis.
3Reliability
If intensive management and early intervention are implemented, then patellar luxation can be prevented, but requires early and accurate detection capability
Solution Approach 1:
The patent enables preliminary detection of joint abnormalities by analyzing video data of animal movement before clinical symptoms fully manifest. The pose estimation system identifies subtle movement pattern deviations that indicate early-stage joint issues, allowing owners to seek preventive treatment and management interventions before the condition progresses to severe patellar luxation.
Solution Approach 2:
The patent replaces the need for specialized veterinary equipment and expertise in early detection with an AI-based pose estimation system that analyzes standard video data. This substitution makes early detection accessible to all pet owners without requiring specialized medical equipment, thereby enabling widespread implementation of early intervention programs for patellar luxation prevention.
Data Source
AI summary
Provided are a service providing system capable of predicting and diagnosing the state of joints, and a method therefor. The method relates to a service providing method capable of predicting and diagnosing the state of joints, which is performed by a server, the method comprising the steps of: extracting image information for each frame from photographed image data about a subject; extracting predictive joint data of the subject by inputting the image information for each frame to a pose estimation model in a plurality of convolutional layers of a convolutional neural network (CNN); and comparing and analyzing the predictive joint data extracted through the pose estimation model on the basis of standard data, and determining the presence of an abnormality in joints of the subject, to generate diagnosed result data, wherein the step of extracting the predictive joint data may comprise the steps of: recognizing the subject as an object by separating the subject from a background in the image information for each frame; setting a bounding box for each frame by detecting a boundary area around the object; extracting predictive coordinate information for each frame about a body part of the object located in the bounding box; extracting an object movement direction of the object located in the bounding box; and extracting the predictive joint data by labeling the predictive coordinate information on the basis of the object movement direction.