Smart Glasses

Problem Statement

This project specifically looks at the use case of a person suffering from Face Blindness i.e. Someone who cannot recognise their closest associates within the wrong context. This impacts their personal and professional lives.

The aim of this project is to create a solution using smart glasses, that uses face recognition to identify known faces and provide the user with cues to help him/her to recognise” the person walking towards them, allowing them to respond appropriately.

The associated app must allow the user to register known faces by uploading images and tagging them or tagging unknown faces captured by the smart glasses.


Technologies Used

For AI vision model:
  • CNNs (2D Convolutions)
  • Haar Cascade and Haar Like Feature Extractors
  • YOLO V4
  • One Shot and Few Shot Learning
  • TensorflowLite
  • Other model optimisation techniques (including quantizations) to enable mobile AI and real-time inference.
  • APIs for model updation (when new data comes in)
For IoT:
  • ESP32
  • OV2640 2MP Cam
  • Wi-Fi
For Application:
  • Flutter
  • Dart
  • AWS