Overview
Mind’s Eye is an assistive smart-glasses system that helps turn visual information into spoken output. The project combines a wearable device, an Android client, and recognition services rather than treating computer vision as a stand-alone model demo.
Challenge
The solution had to coordinate wearable hardware, a mobile client, camera-derived inputs, recognition services, and speech output. Each boundary needed a clear data path so the user received one useful result instead of disconnected component outputs.
My role
I delivered more than 80% of the system. I also built the Egyptian-currency recognition capability from scratch and integrated it into the broader assistive workflow.
What I built
I built the Android client in Java, connected the Arduino-based wearable input to REST recognition flows with Retrofit, and integrated the custom currency-recognition capability with OCR-to-speech behavior.
Architecture
Arduino wearable → Android Java client → REST recognition services → OCR / vision result → text-to-speech
OpenCV supports the computer-vision work, while Tesseract OCR extracts text for spoken playback.
Results
The delivered system brought mobile, embedded, cloud, computer-vision, and speech components into one assistive workflow. No user-count or recognition-accuracy metric is claimed without a verified source.
Lessons
Mind’s Eye showed the value of treating integration as a first-class engineering problem. A recognition feature becomes useful only when device input, service communication, and accessible output work together.