Knowledge-Graph Biased Classification for Noisy Object Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional object recognition systems face challenges in accurately classifying objects, especially in noisy or occluded images, due to their inability to effectively utilize co-occurrence patterns and lack of dynamic learning during training.
Innovation Solution
The implementation of a knowledge-graph biased classification method that applies multiple confidence values to objects, determines a metric based on these values, and uses a knowledge-graph to classify objects when the metric exceeds a threshold, thereby enhancing classification accuracy by leveraging co-occurrence probabilities and adapting to new data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional object recognition systems are used, then the system structure is simple, but the classification accuracy deteriorates in noisy or occluded images
Solution Approach 1:
The system performs preliminary actions by pre-training a knowledge graph with co-occurrence patterns from training data before actual object recognition. This pre-established knowledge base enables the system to make informed predictions about occluded or noisy objects by leveraging learned relationships between objects, thereby improving classification accuracy without requiring complex real-time processing
Solution Approach 2:
The patent introduces a knowledge graph as an intermediary between the input image and the classification output. This knowledge graph stores co-occurrence patterns and acts as a mediator that biases the classification toward more probable objects based on learned relationships, resolving the contradiction by adding a moderate layer of complexity that significantly boosts accuracy in challenging conditions
2Adaptability or versatility
If traditional computational techniques are used, then the system complexity is low, but the ability to utilize co-occurrence patterns deteriorates
Solution Approach 1:
The system performs preliminary action by pre-training the knowledge graph with co-occurrence patterns from training data before actual object recognition. This pre-established knowledge base enables the system to make informed predictions about occluded or noisy objects by leveraging learned relationships, thereby improving classification accuracy without requiring complex real-time processing
Solution Approach 2:
The system implements feedback by using the knowledge graph to provide biased predictions that influence the final classification. The co-occurrence patterns stored in the knowledge graph act as feedback mechanisms that guide the classification toward more probable objects, enabling the system to effectively utilize co-occurrence patterns while maintaining manageable computational complexity
3Measurement precision
If neural networks are used for object recognition, then the classification accuracy improves, but the processing time increases
Solution Approach 1:
The system performs preliminary action by pre-training the knowledge graph with co-occurrence patterns from training data before actual object recognition. This pre-established knowledge base enables the system to make informed predictions about occluded or noisy objects by leveraging learned relationships, thereby improving classification accuracy without requiring complex real-time processing
Solution Approach 2:
The knowledge graph serves itself by storing and utilizing co-occurrence patterns that were learned during training. This self-contained knowledge base allows the system to quickly query probable object classifications without requiring extensive real-time neural network computations, thereby reducing classification latency while maintaining improved accuracy
Data Source
AI summary
A method for classifying an object includes applying multiple confidence values to multiple objects. The method also includes determining a metric based on the multiple confidence values. The method further includes determining a classification of a first object from the multiple objects based on a knowledge-graph when the metric is above a threshold.