Road Obstacle Detection via Visual Saliency and Region Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting road obstacles using machine learning struggle to accurately identify obstacles without prior learning, as they require extensive data collection and cannot account for all possible variations in road obstacles.
Innovation Solution
A road obstacle detection device that segments images into local regions, calculates the probability of a region being an obstacle based on its visual saliency and difference from surrounding regions, and integrates probabilities from multiple segmentation sizes to determine obstacle presence, allowing for accurate detection without pre-learning obstacle data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning methods are used to detect road obstacles, then detection accuracy for known obstacles is improved, but the system cannot detect unknown obstacles that were not included in the learning data
Solution Approach 1:
The image is segmented into multiple local regions, and the system calculates visual saliency for each region independently. This allows the system to detect obstacles of various sizes and types by comparing each local region against its surroundings, rather than requiring pre-trained knowledge of specific obstacle types.
Solution Approach 2:
The system changes the parameter of visual saliency calculation by considering both the probability that a region is not a normal object and the degree of visual difference from surrounding regions. This dual-parameter approach enables detection of diverse obstacles without requiring specific training data for each obstacle type.
2Adaptability or versatility
If extensive learning data is collected to improve obstacle detection coverage, then detection versatility is improved, but data collection time and system complexity increase
Solution Approach 1:
The system performs self-service by automatically calculating visual saliency and obstacle probabilities based on image processing alone, without requiring external data collection or training phases. The method uses inherent visual characteristics of the image to detect obstacles, eliminating the time-consuming data collection process.
3Measurement precision
If the system segments images into multiple local regions and calculates visual saliency, then detection accuracy for various obstacle sizes is improved, but computational complexity increases
Solution Approach 1:
The image is divided into multiple local regions, allowing the system to detect obstacles of different sizes by processing each region independently. This segmentation approach improves detection accuracy for various obstacle sizes while maintaining a systematic processing framework.
Solution Approach 2:
Each local region is processed with specific attention to its visual characteristics and relationship with surrounding regions. This local quality approach allows the system to adapt to different obstacle types and sizes in different parts of the image, improving overall detection accuracy.
Data Source
AI summary
Provided is a road obstacle detection method including: a region segmentation step of segmenting an input image into a plurality of local regions; and a likelihood calculation step of calculating a probability of presence of the road obstacle in a target local region based on a probability that the target local region is not a normal object set in advance and a degree of visual saliency defined by a relationship between a surrounding local region and the target local region, wherein the degree of the visual saliency is calculated to be larger as a probability that the surrounding local region is a road is higher and calculated to be larger as a difference in visual characteristics between the target local region and the surrounding local region is larger.