Entry-Exit Matching System Using Global Re-Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for entry-exit matching in transportation systems, such as manual data collection and Automated Fare Collection systems, are costly, prone to errors, and limited in their ability to assess confidence in matchings, especially in networks with open AFC systems where exit information is not available.
Innovation Solution
A visual person re-identification system that uses machine-learning-based re-identification codes generated from camera images, combining facial features with other attributes like clothing and gait, to automatically match entry and exit events without requiring passenger interaction, employing combinatorial optimization algorithms for global matching and confidence assessment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual surveying and counting methods are used to gather Origin-Destination data, then data collection can be performed without automated systems, but the process becomes costly and prone to errors with small sample sizes
Solution Approach 1:
The patent replaces manual mechanical surveying methods with an automated computer vision system that uses machine learning models to detect and track passengers. Cameras capture images and the system automatically identifies entry and exit events, eliminating the need for manual counting and surveying while improving both accuracy and efficiency.
Solution Approach 2:
The system enables self-service data collection where the transportation system automatically monitors and records its own passenger flow without external human intervention. The machine learning model autonomously processes camera feeds, identifies passengers, and generates Origin-Destination data, allowing the system to serve itself in data collection tasks.
2Loss of information
If Automated Fare Collection systems with smart cards are implemented to track passenger origin and destination, then reliable fare information can be gathered, but exit information remains unavailable in open AFC systems preventing entry-exit matching
Solution Approach 1:
The patent introduces computer vision technology as an intermediary system that bridges the gap in open AFC systems. Instead of relying solely on smart card data, the system uses cameras and machine learning to independently detect and track passengers, creating a parallel information channel that provides exit data even when AFC systems don't capture it.
Solution Approach 2:
The patent segments the passenger tracking function into separate components: the AFC system handles fare collection while the computer vision system independently handles passenger identification and tracking. This segmentation allows the vision system to provide exit information without being constrained by AFC system limitations, enabling entry-exit matching in open systems.
3Measurement precision
If visual soft biometric traits are used to narrow down matching candidates, then the gallery of potential matches can be reduced, but the overall matching strategy remains undefined and confidence assessment is limited
Solution Approach 1:
The patent transforms soft biometric traits into quantitative parameters by using machine learning models to generate numerical representations of passenger features. The system extracts and compares specific parameters such as clothing characteristics, body posture, and gait patterns, converting qualitative visual information into measurable data that enables precise matching with confidence assessment.
Solution Approach 2:
The patent implements a feedback mechanism where the machine learning model continuously refines matching decisions based on comparison results. The system assesses confidence levels for each match and uses this feedback to adjust matching strategies, improving precision while managing system complexity through iterative optimization rather than complex predetermined rules.
4Extent of automation
If re-identification codes are generated from camera images using machine learning, then automated entry-exit matching can be performed without passenger interaction, but the system must handle noisy and occasionally missing input data
Solution Approach 1:
The patent prepares for data quality issues in advance by implementing robustness mechanisms in the machine learning pipeline. The system uses data augmentation, noise filtering, and multiple feature extraction methods to cushion against noisy or missing data before it affects matching accuracy, ensuring reliable automated operation even with suboptimal input quality.
Solution Approach 2:
The patent dynamically adjusts processing parameters based on data quality assessment. When input data is noisy or incomplete, the system changes parameters such as feature weighting, matching thresholds, and confidence requirements to maintain reliable automated matching despite variable data quality conditions.
Data Source
AI summary
Examples relate to a concept for an entry-exit matching system, and in particular to an evaluation device, a method and a computer program for person re-identification for entry-exit matching in a transportation system. The evaluation device comprises processing circuitry configured to obtain a plurality of re-identification codes. Each re-identification code represents a person being recorded by at least one camera when entering or exiting at least a section of the transportation system. The processing circuitry is configured to match the plurality of re-identification codes using a global matching scheme to obtain a plurality of matched pairs of re-identification codes, such that each matched pair of re-identification codes comprises a re-identification code of a person entering and a re-identification code of a person exiting. The global matching scheme is based on reducing an overall distance between the re-identification codes of the matched pairs of re-identification codes over the plurality of matched pairs of re-identification codes. The processing circuitry is configured to determine points of entry and exit for the plurality of matched pairs of re-identification codes.