Semantic Image Segmentation for Object Recognition Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional generic object recognition methods using Spatial Pyramid Matching (SPM) are influenced by the position, size, and background clutter of objects, leading to decreased recognition accuracy.
Innovation Solution
An image recognition device that segments input images based on semantic maps, generating histograms that reflect the content information of the image, thereby reducing the influence of object position, size, and background clutter, and improving recognition accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If equal segmentation of input image is used, then device complexity is reduced, but recognition accuracy deteriorates due to influence of object position, size and background clutter
Solution Approach 1:
The patent applies segmentation by dividing the input image into multiple regions based on semantic information rather than equal geometric division. The segmenting unit segments the input image into a plurality of regions in accordance with meanings extracted from content of the input image, creating semantically meaningful regions that improve recognition accuracy while managing complexity through automated semantic analysis.
Solution Approach 2:
The patent implements local quality by making different regions of the image have different processing characteristics based on their semantic content. Each segmented region is processed individually with feature data computation tailored to its semantic meaning, allowing the system to adapt to local variations in object position, size, and background clutter rather than applying uniform processing.
2Measurement precision
If semantic-based segmentation is used, then recognition accuracy is improved, but device complexity increases due to semantic map generation and content analysis
Solution Approach 1:
The patent applies preliminary action by generating semantic maps and extracting content information before the main segmentation and recognition processes. The segmenting unit first segments the input image into regions based on pre-computed semantic meanings, and the generating unit then computes feature data for each region. This preliminary semantic analysis enables subsequent processing to focus on semantically relevant features, improving accuracy while managing complexity through staged processing.
3Productivity
If conventional SPM histogram computation is used, then processing speed is maintained, but recognition accuracy deteriorates due to positional and size influence
Solution Approach 1:
The patent implements dynamics by making the segmentation and feature computation adaptive to the content of each image rather than using fixed geometric divisions. The segmenting unit dynamically segments the input image based on extracted semantic content, and the generating unit dynamically computes feature data for each semantically-defined region. This dynamic adaptation allows the system to maintain processing efficiency while improving recognition accuracy by focusing on semantically relevant features.
Data Source
AI summary
An image recognition device that improves the accuracy of generic object recognition compared with conventional technologies by reducing the influence of the position, size, background clutter and the like of an object that is targeted to be recognized in the input image by the generic object recognition. The image recognition device performs a generic object recognition and includes: a segmenting unit configured to segment an input image into a plurality of regions in accordance with meanings extracted from content of the input image; a generating unit configured to compute feature data for each of the plurality of regions and generate feature data of the input image reflecting the computed feature data; and a checking unit configured to check whether or not a recognition-target object is present in the input image in accordance with the feature data of the input image.