Object Image Segmentation for Accurate Product Color Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing techniques struggle to accurately extract product colors for similarity searches, leading to inaccurate color representation and search results.
Innovation Solution
An information processing apparatus utilizing a YOLO learning model for object detection and an FCN learning model for semantic segmentation to determine pixel colors, combined with a feature vector concatenation process for high-accuracy color extraction and similarity search.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If simple foreground-background separation is used to extract product area, then processing speed is improved, but color extraction accuracy deteriorates
Solution Approach 1:
The patent divides the image processing into two distinct stages: first, a learning model (such as YOLO or FCN) segments the product area from the background to obtain a mask image; second, color extraction is performed only on the pixels within the segmented product area. This segmentation approach enables accurate color extraction by isolating the product region, while the use of efficient learning models maintains reasonable processing speed.
2Measurement precision
If learning models (YOLO, FCN) are used for accurate object detection and segmentation, then color extraction accuracy is improved, but device complexity increases
Solution Approach 1:
The patent employs specialized learning models (YOLO for object detection, FCN for semantic segmentation) that are designed to process images efficiently. These models segment the product area with high accuracy, enabling precise color extraction. The complexity is managed by using established architectures that balance accuracy and computational requirements.
Solution Approach 2:
The patent introduces a mask image as an intermediary between the original product image and the color extraction process. The learning model generates the mask image that identifies product pixels, which then serves as a guide for accurate color extraction. This intermediary structure simplifies the overall process by separating detection from color analysis.
3Device complexity
If simple foreground-background separation is used, then device complexity is reduced, but search accuracy deteriorates
Solution Approach 1:
The patent uses learning models to segment the product area with high precision, creating an accurate mask that defines the product boundaries. This segmentation ensures that color extraction is performed only on relevant product pixels, significantly improving search accuracy for similar products while maintaining manageable processing complexity through automated model-based approaches.
Data Source
AI summary
An information processing apparatus (1) includes: an acquisition unit (101) configured to acquire an object image that contains one or more objects; a first inference unit (105) configured to infer one or more rectangular areas that respectively surround the one or more objects in the object image, and a type of each of the one or more objects, by applying the object image to a first learning model; a second inference unit (105) configured to infer an area of a target object in the object image by applying the one or more rectangular areas and the type of each of the one or more rectangular areas to a second learning model; and an extraction unit (105) configured to perform color determination on each pixel in the area of the target object and extract colors of the target object.