Dual-Stage Traffic Light Recognition for Visually Impaired
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing walking support systems for visually impaired individuals face challenges in accurately recognizing traffic lights, leading to unreliable notifications for stopping and crossing, which can compromise safety.
Innovation Solution
A walking support system equipped with an image acquisition unit, a traffic light first recognition unit, a traffic light second recognition unit, and a determination unit, utilizing learned models and feature amounts to accurately recognize traffic lights and provide reliable stop and cross notifications, with re-recognition processes to enhance reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single recognition method is used for traffic light detection, then the device complexity is reduced, but the recognition accuracy becomes insufficient
Solution Approach 1:
The traffic light recognition system is segmented into two independent recognition units: a first recognition unit that performs initial detection and a second recognition unit that performs verification. Each unit uses different recognition approaches, and their results are combined through a determination unit to achieve high accuracy while maintaining modular system architecture.
Solution Approach 2:
A determination unit is introduced as an intermediary component that receives recognition results from both the first and second recognition units. This intermediary processes and integrates the results from multiple recognition methods, resolving the contradiction by adding a coordination layer that improves accuracy without requiring complete system redesign.
2Reliability
If multiple recognition units are used to improve recognition accuracy, then the reliability is improved, but the device complexity increases
Solution Approach 1:
The system merges the outputs of multiple recognition units through a determination unit that integrates their results. This merging approach allows the system to benefit from multiple independent recognition processes while managing complexity through a unified determination mechanism that consolidates the multiple inputs into a single reliable output.
Solution Approach 2:
The system implements feedback mechanisms where the determination unit processes results from multiple recognition units and uses this information to improve overall system reliability. The feedback loop allows the system to learn from and correct recognition errors, enhancing reliability without proportionally increasing complexity.
3Measurement precision
If re-recognition processing is performed to enhance reliability, then the notification accuracy is improved, but the processing time increases
Solution Approach 1:
The first recognition unit performs preliminary detection and identification of traffic lights before the second recognition unit conducts verification. This preliminary action allows the system to quickly identify potential targets and then apply more thorough verification only where needed, reducing overall processing time while maintaining high accuracy through the two-stage approach.
Solution Approach 2:
The determination unit skips redundant processing by intelligently evaluating results from the first and second recognition units. When both units agree on a recognition result, the system can proceed without additional verification steps, effectively skipping unnecessary processing time while maintaining high reliability through the dual-recognition architecture.
Data Source
AI summary
It is determined whether a traffic light is recognized in an image based on each of a result of a recognition operation by a traffic light first recognition unit (a recognition operation of the traffic light for an image acquired by a camera using a learned model based on pre-annotated data) and a result of a recognition operation by a traffic light second recognition unit (a recognition operation of the traffic light for an image acquired by the camera based on a feature amount of the traffic light). This makes it possible to sufficiently obtain the recognition accuracy of the traffic light, and to appropriately perform an instruction to a pedestrian according to the state of the traffic light.