Image Recognition Using Exclusive Classifier for Robust Object Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image recognition technologies lack the robustness to accurately distinguish objects that are predicted not to coexist in the same image, leading to erroneous results.
Innovation Solution
An image recognition apparatus that calculates existence probabilities of candidate objects using feature information and exclusive relationship information to adjust these probabilities, ensuring that objects predicted not to coexist do not have high coexistence probabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If image recognition uses context information and co-occurrence relationships to improve robustness, then recognition accuracy is improved, but erroneous results occur when objects predicted not to coexist are assigned high existence probabilities
Solution Approach 1:
The patent applies preliminary anti-action by pre-defining exclusive object sets containing objects predicted not to coexist in the same image. During recognition, when candidate objects are identified, the system proactively checks against these exclusive sets and adjusts existence probabilities downward for objects that should not coexist, preventing erroneous high probability assignments before final recognition results are produced.
Solution Approach 2:
The patent implements feedback by using exclusive relationship information to continuously adjust existence probabilities during the recognition process. The system monitors the co-occurrence patterns of detected objects and applies corrective adjustments based on pre-established exclusive object relationships, creating a feedback loop that refines recognition accuracy by suppressing improbable co-occurrences.
2Adaptability or versatility
If the system calculates existence probabilities for multiple candidate objects, then comprehensive object detection is achieved, but the computational complexity and processing time increase
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing exclusive object sets containing objects predicted not to coexist during an offline training phase. This preliminary preparation of relationship information allows the online recognition system to efficiently adjust existence probabilities using pre-computed data, reducing real-time computational complexity while maintaining comprehensive detection capabilities.
Data Source
AI summary
The present invention provides an image recognition apparatus with enhanced performance and robustness.In an image recognition apparatus 1, an image classification information accumulating unit 20 stores therein feature information defining visual features of various objects obtained through a learning process. For classification of input images, an image feature obtaining unit 18 extracts descriptors indicating features from each input image, image vocabularies corresponding to the descriptors are voted, and a classifying unit 19 calculates existence probabilities of one or more candidate objects, based on the result of the voting. According to the existence probabilities, objects contained in the image is identified. Through the calculation, the existence probabilities are adjusted by an exclusive classifier, based on exclusive relationship information defining exclusive object sets each containing different objects (object labels) predicted not to coexist in a same image.