Object Detection Using Change Region Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image recognition technologies, particularly deep neural networks, face challenges in accurately detecting objects like people whose shape changes, especially in complex backgrounds, leading to deteriorated detection accuracy.
Innovation Solution
An information processing apparatus that acquires an input image, extracts a change region image, and uses a convolutional neural network to combine this with the original image, suppressing background influence by inputting a change region image with background differences, enabling accurate detection of object positions even in complex backgrounds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If deep neural network is used for object detection, then recognition accuracy is improved, but detection accuracy deteriorates when object shape changes and background is complicated
Solution Approach 1:
The patent segments the image processing into two distinct parts: (1) extracting a change region image that highlights only the moving object by removing static background, and (2) inputting this change region image to the deep neural network for detection. This segmentation allows the network to focus on the object of interest without being confused by complex backgrounds, thereby maintaining high detection accuracy even when object shape changes.
Solution Approach 2:
The patent extracts the change region image by subtracting the background image from the input image, effectively taking out the moving object from the complex background. This extracted change region image is then input to the deep neural network, enabling accurate detection by eliminating the harmful influence of complicated backgrounds while preserving the object's shape information.
2Reliability
If change region image extraction is performed, then background influence is suppressed, but processing complexity increases
Solution Approach 1:
The system performs self-service by automatically extracting the change region image through simple background subtraction operations. The processing apparatus itself generates the change region image by comparing the input image with the background image, eliminating the need for external preprocessing tools or manual intervention, thereby reducing overall system complexity while maintaining high detection accuracy.
Data Source
AI summary
An information processing apparatus includes an acquiring unit to acquire an input image including an object, an extraction unit to extract a region that has been changed from a background image in the input image, as a changing region, a generating unit to generate a change region image in which a pixel value of each pixel represents whether or not a corresponding pixel in the input image belongs to the changing region, an extraction unit to extract a plurality of feature images from the input image as an output of a neural network by inputting a combination of the input image and the change region image to the neural network, and a detection unit to detect a position of the object from the plurality of feature images.