Class of 2026Sammamish, Washington
- "A Computer Vision Based System to Make Street Crossings Safer for the Visually Impaired" with mentor Hieu (Working project)
A Computer Vision Based System to Make Street Crossings Safer for the Visually Impaired
Started Aug. 29, 2023
Abstract or project description
This paper/project introduces a novel system designed to enhance the safety of visually impaired individuals during street crossings. The proposed system utilizes a handheld device to detect pedestrian walk signals and employs object detection to monitor oncoming vehicles. The handheld device consists of a small computer, speaker, camera and battery for real time edge detection. The system starts by detecting any walk signs on pedestrian street lights. After confirming no cars are in the way, the user is told that it is safe to walk. While the user is crossing the street, the device continues to check for any incoming cars using a combination of object detection and depth estimation. The device will check between frames to see if the cars are moving and will see how far away each car is. If the device determines that a collision with a pedestrian is imminent, the device will make an audio warning to user telling them to back up, thereby saving their life.