Android application for AI assistant designed specifically to empower individuals with visual impairments, which provides users with a seamless and supportive experience in navigating their surroundings.
Untitled.video.-.Made.with.Clipchamp.mp4
An Android phone sends the photo and the question to the FastAPI backend. It appends the question to the chat log. Then, a classifier based on ChatGPT analyzes whether it needs visual context. If so, it uses BLIP via the Replicate.com API and processes it. If not, it answers the question as it is. To see the implementation, please refer to the backend repository.
- image capturing
- image processing
- speech recognition
Language: Kotlin
Framework: Native Android UI
Architecture: MVVM
Speech Recognition | Responding | Answer is displayed and voiced |
---|---|---|
This application was developed for the GDSC Solution Challenge 2023. You can watch our demo video here