False Negative Detection via Historical Track Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In camera systems, identifying images with false negative object detections is a cumbersome task requiring manual processing, prone to errors, as these images are not associated with any object detections, leading to inefficient filtering and potential misleading results.
Innovation Solution
A method that compares unmatched tracks with reference tracks to determine if an object detection is likely missing, using historic tracks grouped by object class, allowing for resource-efficient filtering of images with potential false negative object detections, reducing manual work and errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual processing and review is used to find images with false negative object detections, then comprehensive identification of missed detections is achieved, but the process becomes cumbersome and time-consuming
Solution Approach 1:
The system performs preliminary automated filtering by comparing unmatched tracks with reference tracks before manual review. This pre-processing step identifies likely false negatives using historical data and similarity metrics, reducing the volume of images requiring manual verification while maintaining high identification accuracy.
Solution Approach 2:
Reference tracks serve as an intermediary between raw unmatched tracks and final false negative identification. By introducing this intermediate layer of historical track data, the system automates the filtering process while preserving the reliability that would otherwise require extensive manual review.
2Reliability
If manual filtering of images is performed, then images with false negatives can be identified, but human errors lead to faulty or misleading results
Solution Approach 1:
The system uses its own historical track data to automatically evaluate and filter unmatched tracks. By making the system self-sufficient in identifying false negatives through automated comparison with reference tracks, human intervention is minimized, eliminating human errors while maintaining consistent and reliable results.
Solution Approach 2:
The system incorporates feedback loops where identified false negatives and their characteristics are fed back into the reference track database. This continuous learning mechanism improves the accuracy of future false negative identification, systematically reducing errors over time without relying on manual correction.
3Reliability
If all images are processed to ensure complete identification of false negatives, then no missed detections are overlooked, but resource consumption increases
Solution Approach 1:
The system applies partial action by processing only a subset of images through automated comparison with reference tracks. By selectively evaluating unmatched tracks based on similarity metrics rather than processing all images exhaustively, the system maintains high detection completeness while significantly reducing computational resource consumption.
Solution Approach 2:
The image processing task is segmented into distinct stages: automated filtering using reference track comparison, and selective manual review of only those cases that meet specific criteria. This segmentation allows the system to achieve comprehensive false negative identification while minimizing overall resource usage by automating the bulk of the work.
Data Source
AI summary
A method for determining images plausible to have a false negative object detection comprises providing a group of historic trajectories, wherein each historic trajectory comprises a reference track that represents one or more historic tracks and comprises an object class of historic object detections that belong to the one or more historic tracks; performing tracking; performing object detection; for a determined track that does not match any determined object detection, comparing the determined track with reference tracks of historic trajectories for identifying a matching reference track; upon identifying a matching reference track, defining images of the determined track as being plausible to have a false negative object detection for the object class of the historic trajectory comprising the matching reference track; and upon not identifying a matching reference track, defining the determined track as a false positive track.