Purpose
Over 2 billion people worldwide are vision impaired. I am vision impaired and find it difficult to gauge a car’s speed or distance. I have been involved in near road accidents when cars approached me on my blind side.
Unable to find a suitable device, I began developing VIPMOD: Vision Impaired Person’s Moving Object Detector.
The aim of my project is to design an app which detects fast moving objects, so that vision impaired people can live safer and more independent lives.
Method
This project evolved through six prototypes. The first three prototypes used Micro:bits.
Prototype 4 uses GPS technology.
Prototype 5 is an app which uses TensorFlow.js Image Classification model to detect oncoming vehicles and other objects. The app displays the object's name, vibrates and issues a text-to-speech warning.
Prototype 6 is an application that estimates the speed of oncoming objects. It uses a YOLOv8 model to analyse input footage.
Results
Prototype 5 was tested in a controlled setting, with 95% accuracy (316 trials).
Vision Ireland will be testing VIPMOD in the WayFinding Centre - an indoor environment replicating the real-world experience of using public transport for vision impaired people.
Conclusion
From the data collected, I conclude that VIPMOD has the potential to improve road safety for all road users, particularly for vision impaired.
I am consulting with Dr Anna Zamansky and her team from Haifa University, to further develop VIPMOD.
I aim to develop the VIPMOD app to promote safety and independent travel for all.
|