A Computer Vision Based System to Make Street Crossings Safer for the Visually Impaired

Overview

Crosswalks are dangerous places with high risks of pedestrians coming into contact with cars. Unfortunately, this problem is especially true for visually impaired pedestrians, as they cannot see when a distracted driver drives past the red light, potentially causing them a fatal injury. Computer based solutions to make crosswalks safer focus more on detecting if the crosswalk sign is on and making sure the user knows the path to the other side of the street. However, these solutions do not take into account that a distracted driver might run past the red light and potentially cause any injury to the user. We aim to create a system that uses a camera stream and detects if there are any incoming vehicles while a visually impaired user is crossing a crosswalk and notifies them of any potential hazards. This system detects the vehicles using object detection and tracks their distance to the user using depth detection. Our system is embedded on a Raspberry Pi and an Intel Neural Compute Stick 2. The models and logic are tested against real world data collected from neighborhoods around Seattle, Wash ington. The results indicate a high precision of 96%, which allows for continuous advancement and deployment in future research.

Competition history

• JSHS 2024 Category not listed

Resources

• National JSHS abstract book ↗

Related projects

Closest projects by meaning, across every fair and year in the corpus.