Multi-Model AI Image Classification Pipeline
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automatic labeling methods for images are inefficient due to low accuracy, requiring separate pre-processing and manual classification, which increases operating time and necessitates verification by inspectors, and do not provide accurate results when images are incorrectly classified for unsuitable models.
Innovation Solution
An automatic image classification and processing method using a continuous processing structure of multiple AI models, where image data is processed through feature extraction models to generate feature values, which are then input into classification models to determine suitable labeling types, enabling simultaneous labeling type classification and automatic processing, improving accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If a single artificial intelligence model is used for automatic labeling, then the device complexity is low, but the manufacturing precision (labeling accuracy) is insufficient
Solution Approach 1:
The patent divides the labeling system into multiple specialized AI models, each trained for specific labeling tasks (e.g., object detection, segmentation, keypoint detection). This segmentation allows each model to excel at its specific function, improving overall labeling accuracy while maintaining manageable complexity through modular architecture.
Solution Approach 2:
The patent creates a universal labeling system where multiple AI models can process different types of labeling tasks. The system architecture allows a single framework to handle various labeling requirements (bounding boxes, segmentation masks, keypoints) through multiple specialized models, achieving multi-functionality without requiring a completely separate system for each task.
2Productivity
If manual classification and verification processes are used, then the reliability of labeling is improved, but the productivity (processing speed) deteriorates
Solution Approach 1:
The patent implements self-service mechanisms where the AI models automatically perform classification and verification of labeling results. The system includes automatic quality assessment capabilities where models evaluate their own output confidence levels and can perform self-verification, reducing the need for manual intervention while maintaining high reliability through automated quality control.
3Loss of time
If separate pre-processing and classification steps are used, then the measurement precision (classification accuracy) is improved, but the loss of time (operating time) increases
Solution Approach 1:
The patent merges the pre-processing and classification steps into an integrated pipeline where feature extraction and classification occur in a continuous flow. The system combines multiple AI models in a sequential architecture where the output of one model directly feeds into the next, eliminating intermediate manual intervention steps and reducing total processing time while maintaining classification accuracy through the coordinated operation of specialized models.
Data Source
AI summary
Disclosed is an automatic image classification and processing method based on the continuous processing structure of multiple artificial intelligence models. An automatic image classification and processing method based on a continuous processing structure of multiple artificial intelligence models includes receiving image data, generating a first feature extraction value by inputting the image data into a first feature extraction model among feature extraction models, generating a second feature extraction value by inputting the image data into a second feature extraction model among the feature extraction models, and determining a classification value of the image data by inputting the first and second feature extraction values into a classification model.