Occluded Object Detection in Autonomous Sensor Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional systems fail to accurately label images generated by machines navigating environments due to occlusions, leading to incorrect ground truth data and increased manual effort for quality checking.
Innovation Solution
The system detects occluded objects in images or sensor data representations by processing images to determine classifications and using maps to project labels, combining techniques to indicate occluded portions of objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional systems use map labels to generate image labels, then the labeling process is simple and fast, but the ground truth data becomes inaccurate when objects are occluded
Solution Approach 1:
The system performs preliminary occlusion detection by comparing map labels with detected objects before final label generation. This preliminary check identifies occluded regions where map labels would be incorrect, allowing the system to flag or correct these labels before they are used for training, thus preventing propagation of errors while maintaining overall system efficiency
Solution Approach 2:
The patent introduces an intermediary occlusion detection mechanism that acts as a mediator between map-based labeling and final ground truth generation. This intermediary layer compares expected map labels with actual detected objects and identifies discrepancies caused by occlusions, allowing for selective correction or flagging of problematic labels without completely replacing the efficient map-based approach
2Measurement precision
If conventional systems manually quality check labels, then ground truth accuracy improves, but time consumption and manual effort increase
Solution Approach 1:
The system performs self-quality-checking by automatically detecting occlusions through comparison of map labels with detected objects. The occlusion detection mechanism serves as a self-service quality control layer that identifies and flags problematic labels without requiring human intervention, allowing the system to maintain high accuracy while operating autonomously
Solution Approach 2:
The patent implements a feedback mechanism where occlusion detection results are fed back into the labeling process. When occlusions are detected, the system automatically adjusts or flags the corresponding labels, creating a closed-loop quality control system that continuously improves ground truth accuracy without adding manual review steps
3Productivity
If conventional systems generate labels without occlusion detection, then productivity is high, but the reliability of training data decreases
Solution Approach 1:
The system applies partial occlusion detection only to regions where map labels are projected, rather than performing complete scene analysis. This selective approach focuses computational resources on critical labeling areas, maintaining high productivity while sufficiently improving training data reliability by catching the most problematic occluded labels
Data Source
AI summary
In various examples, detecting occluded objects within images or other sensor data representations for autonomous or semi-autonomous systems and applications is described herein. Systems and methods described herein may determine when objects are occluded at portions of images using various techniques. For example, an image may be processed in order to determine classifications associated with objects depicted by the image and, the classifications, along with labels that are projected on the image using a map, may then be used to determine whether one or more of the objects are occluded in the image. For another example, a map may be used to determine first distances to points within an environment and a point cloud may be used to determine second distances to the points within the environment. The distances may then be used to determine whether one or more objects are occluded within the image.