Sound generating systems are installed only 7.2% of crosswalks in South Korea and 46.9% of those lack of regular maintenance, possibility of safe crossing is remarkably low for the visually impaired.
This research aims to establish a real-time robust visual object detection algorithm and develop a pedestrian crossing aid device based on that algorithm.
Pedestrian signal detection and tracking
Detection module is composed of Color-based segmentation and machine learning recognition.
Continuously Adaptive Mean Shift is used for tracking module.
Cascade color segmentation is used for speed and reliability.
Window setting method is used to shorten the processing time.
System input: presence of the pedestrian signal and its color
System output: guiding voice
Total Ground Truth: 1850 images
Detection Rate = 0.7735 (FAR = 0.0124, FPR = 0.2189, Specificity = 0)
Color Interpretation Accuracy = 0.9622
Overall system accuracy = 0.7443