Object Location Detection via Part Segmentation and Homography
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing augmented reality (AR) applications on portable user devices face challenges in real-time object detection due to the computational complexity of Convolutional Neural Network (CNN) and part-based models, which are resource-intensive and unsuitable for devices with limited processing capabilities.
Innovation Solution
A method that employs simpler object detectors to detect object parts and uses geometric models and homography to estimate object locations, reducing computational complexity by leveraging detected object parts to identify objects in a picture.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If CNN-based object detection solutions are used to improve detection accuracy, then measurement precision is improved, but device complexity and processing requirements increase making them unsuitable for portable devices
Solution Approach 1:
The patent divides the object detection task into two stages: first detecting individual object parts (segments) using simple detectors, then combining these parts according to a geometric model to identify the complete object. This segmentation approach allows each stage to use computationally efficient methods while achieving accurate object-level detection, resolving the contradiction between accuracy and processing complexity for portable devices.
2Device complexity
If part-based object detection models are used to reduce processing requirements, then device complexity is reduced, but detection precision and accuracy deteriorate
Solution Approach 1:
The patent performs preliminary detection of object parts using simple, computationally efficient detectors before performing the more complex task of assembling these parts into complete object detections using geometric models. This preliminary action at the part level enables the system to use less complex processing while maintaining accuracy through the structured combination of detected parts according to known geometric relationships.
3Measurement precision
If complex geometric models and homography estimation are used to improve object location precision, then measurement precision is improved, but processing time and computational energy increase
Solution Approach 1:
The patent applies geometric models and homography estimation locally to each detected object part and its corresponding region in the image, rather than performing global optimization across the entire image. This local approach to geometric verification allows the system to achieve precise object location determination while significantly reducing computational time and energy requirements compared to global optimization methods.
Data Source
AI summary
Object parts (20, 21, 22, 23, 24) are detected in a picture using object detector(s) (3) and part location representations (40, 42, 43, 44) are generated for the detected object parts (20, 22, 23, 24). The size of an object (10) comprising object parts (20, 21, 22, 23, 24) is estimated based on a geometric model and the part location representations (40, 42, 43, 44). Search locations (51) in the picture for a search window (52) having a size based on the estimated size are determined based on the part location representations (40, 42, 43, 44). The search locations (51) are then processed by identifying any detected object part (20, 22, 23) that is within the search window (52) positioned at the search location (51). A homography is estimated by minimizing an error between mapped object part(s) from the geometric model and the identified detected object part(s) (20, 22, 23). If the error is smaller than a threshold value, an object location representation is determined for the object (10).