Image Evaluation System Matrix Composition Subject Bias
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image evaluation systems tend to be biased towards selecting images with faces, leading to an imbalance in the selection of scenic and still-life images when generating electronic albums, as they prioritize images with high face detection scores over other types of content.
Innovation Solution
An image evaluation system that calculates composition and subject evaluation values based on frequency of appearance, using a matrix product to evaluate images without bias towards face-inclusive images, allowing for the selection of scenic and still-life images by considering composition and subject types independently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If image evaluation prioritizes images with faces, then images with faces are selected more frequently, but scenic images and still-life images are underrepresented in the selection
Solution Approach 1:
The image evaluation system segments the evaluation criteria into multiple independent components: face detection results, composition evaluation, and subject evaluation. Each component is calculated separately and then integrated through a composite scoring formula, allowing balanced consideration of different image types without overwhelming bias toward any single criterion
Solution Approach 2:
The system dynamically adjusts evaluation parameters by introducing composition evaluation values and subject evaluation values that can compensate for face detection scores. By changing the parameter structure from single-criterion (face detection) to multi-criterion (face + composition + subject), the system achieves better balance in image selection
2Ease of operation
If conventional image evaluation uses only face detection scores, then the evaluation process is simple, but the selection becomes one-sided towards images with faces
Solution Approach 1:
The evaluation process is segmented into distinct modules: face detection module, composition analysis module, and subject identification module. Each module operates independently with its own calculation logic, maintaining operational simplicity while improving assessment accuracy through modular design
Solution Approach 2:
The evaluation system is designed with multi-functionality to handle different image types universally. The composite scoring formula can evaluate portraits, landscapes, and still-life images equally well by considering multiple factors (face detection, composition, subject) that are relevant across all image categories
3Productivity
If images are selected based on high face detection scores, then face-inclusive images are prioritized, but rare compositions and subjects are overlooked
Solution Approach 1:
The system performs preliminary evaluation of composition and subject characteristics before final image selection. By pre-calculating composition evaluation values and subject evaluation values for all candidate images, the system preserves information about rare compositions and subjects, allowing them to be considered in the final composite scoring rather than being overlooked
Solution Approach 2:
The composite scoring formula acts as an intermediary that balances face detection results with composition and subject evaluations. This intermediary mechanism ensures that images with rare compositions or subjects can still achieve high overall scores even if their face detection scores are moderate, preventing loss of diversity information
Data Source
AI summary
Provided are an image evaluation system, as well as an image evaluation method, and recording medium storing an image evaluation program, in which when an image is selected from among multiple images, images are evaluated in such a manner that an image such as a scenic image or still-life image will be selected with little image bias. To achieve this, an individual image composition/subject matrix is found, the matrix comprising types of composition and types of subject included in an image. A total image composition/subject matrix is obtained by adding individual image composition/subject matrices regarding respective ones of multiple images. A composition evaluation matrix and a subject evaluation matrix are found from the total image composition/subject matrix. Image evaluation values regarding respective ones of the multiple images are calculated from a matrix product of the individual image composition/subject matrix, composition evaluation matrix and subject evaluation matrix.