Road Obstacle Detection Using Semantic Label Distribution Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing road obstacle detection systems face challenges in improving detection accuracy due to failures in assigning semantic labels, leading to incorrect obstacle identification.
Innovation Solution
A road obstacle detection device that uses pre-learned identifiers to associate semantic labels with each pixel of an image, estimates the statistical distribution of semantic labels in a region of interest from surrounding areas, and calculates the likelihood of an object being a road obstacle using statistical distributions from both the region of interest and its peripheral area.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If semantic labels are assigned to each pixel using a pre-learned identifier, then the detection process can be simplified and executed efficiently, but detection accuracy deteriorates due to incorrect label assignments when semantic labels are also assigned to obstacles
Solution Approach 1:
The image is divided into a region of interest and a peripheral region, allowing different processing approaches for each area. The region of interest undergoes statistical distribution estimation from the peripheral region, while other areas use direct semantic labeling, thus resolving the contradiction between processing efficiency and detection accuracy.
Solution Approach 2:
A statistical distribution estimation mechanism acts as an intermediary between the peripheral region's semantic labels and the region of interest's obstacle detection. This intermediary process refines the semantic label assignments by considering contextual information from surrounding areas, improving accuracy without sacrificing overall processing efficiency.
2Measurement precision
If statistical distribution estimation is performed for the region of interest from the peripheral region, then detection accuracy is improved by reducing incorrect semantic label assignments, but device complexity increases
Solution Approach 1:
Different processing qualities are applied to different regions: the region of interest receives enhanced statistical distribution estimation for higher accuracy, while other regions use standard semantic labeling. This localized approach improves detection accuracy where needed without unnecessarily increasing complexity across the entire system.
Solution Approach 2:
Statistical distribution estimation is applied partially only to the region of interest rather than the entire image. This partial action provides the necessary accuracy improvement for obstacle detection while avoiding the excessive computational complexity that would result from applying the same process to all image regions.
Data Source
AI summary
A road obstacle detection device which uses a pre-learned first identifier to associate a semantic label with each pixel of an image, uses a pre-learned second identifier to estimate a statistical distribution of a semantic label of a predetermined region of interest of the image from a statistical distribution of a semantic label of a peripheral region that surrounds the region of interest, and uses the statistical distribution of the semantic label associated with the region of interest and the statistical distribution of the semantic label estimated for the region of interest to estimate a likelihood that an object is a road obstacle.