Object Detection via Multi-Aspect Ratio Image Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional object detection methods face challenges in accurately detecting small objects within images due to differences in aspect ratios, leading to reduced accuracy and increased buffer size requirements.
Innovation Solution
The method involves capturing and resizing multiple detection portions from an input image with different aspect ratios to compute coordinates and object ranges, preventing over-shrinking of target objects and reducing buffer size, while using a 2-crop approach to maintain object size and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If the input image is scaled to meet the size and aspect ratio of the detection image, then the detection image can be generated, but blank areas appear and object size is reduced
Solution Approach 1:
The input image is divided into multiple sub-images (first sub-image and second sub-image) with different aspect ratios. Each sub-image is processed separately through scaling and cropping operations, then merged to form the final detection image. This segmentation allows preserving original object dimensions while filling the detection image completely without blank areas.
2Adaptability or versatility
If the input image is scaled down to fit the detection image dimensions, then the aspect ratio mismatch is resolved, but the object size becomes smaller and harder to detect
Solution Approach 1:
Instead of scaling the entire input image, the method segments it into sub-images that can be independently processed. Each sub-image maintains better proportionality to the detection image dimensions, reducing excessive scaling and preserving object size for accurate detection.
Solution Approach 2:
Different regions of the input image are processed with different scaling and cropping operations tailored to their specific aspect ratios. This local processing ensures that objects in different regions maintain appropriate sizes relative to the detection image, improving overall detection accuracy.
3Measurement precision
If multiple detection portions are captured with different aspect ratios, then object detection accuracy is improved, but the processing complexity increases
Solution Approach 1:
The input image is divided into a fixed number of sub-images (typically two) with predetermined aspect ratios. This structured segmentation simplifies the processing complexity compared to arbitrary multi-scale processing, while still achieving improved detection accuracy through targeted region processing.
Data Source
AI summary
An object detection method, for detecting a target object, comprising: capturing at least two detection portions with a first aspect ratio from an input image with a second aspect ratio; confirming whether any object is detected in each of the detection portions and obtaining corresponding boundary boxes for detected objects; and wherein the first aspect ratio is different to the second aspect ratio.