WeTry
Vision for the blind.
Core trick
Camera frames go through OCR and object detection into an LLM, and the answer is spoken in Hindi, English, or Kannada via TTS.
Description
Vision for the blind.
We have build something for long term vision dedicating it to my little cousin sister who is blind. this is for her. Help us get to right people and places to make this a reality. Google phones and apps for EVERYONE
README excerpt
# Netra — AI Vision Companion > Native Android · Kotlin + Jetpack Compose · Offline-first Netra (दृष्टि, "vision") helps visually impaired users understand their surroundings in real time. Point the camera at anything — the app describes what it sees out loud in **Hindi, English, or Kannada**. ## Demo **Camera → OCR + Object Detection → LLM → Sarvam AI TTS** | Feature | How | |---|---| | Scan scene | Tap button or press **Vol↓** | | Ask question | Hold **Vol↑** (push to talk) or tap mic | | Switch language | Tap language button (HI / EN / KN) | | Surveillance | Haptic alert when scene changes significantly | ## AI Stack | Tier | Model | When | |---|---|---| | Cloud vision | Gemini 3.1 Flash | Online + image | | Cloud text (free) | DeepSeek V3 | Online fallback | | Local network | Ollama (any model) | Same WiFi as laptop | | On-device | Gemma 3 1B INT4 | Fully offline | **TTS:** Sarvam AI `bulbul:v1` (online) → Android TTS (offline) **STT:** Google Speech Recognition (online) → offline fallback ## Setup ### 1. Clone & open in Android Studio ```bash git clone <github link> ``` Open in Android Studio Jellyfish or later. ### 2. Add API keys to `local.properties` ```properties sdk.dir=/path/to/Android/Sdk OPENROUTER_API_KEY=sk-or-v1-... SARVAM_API_KEY=sk_... ``` Get keys from: - OpenRouter: https://openrouter.ai/keys - Sarvam AI: https://dashboard.sarvam.ai ### 3. Build & run ```bash ./gradlew assembleDebug adb install app/build/outputs/apk/debug/app-debug.apk `