Skip to content

Repository files navigation

👁️ SeeForMe

Empowering independence for visually impaired individuals through real-time, multilingual scene understanding with audio feedback - now 100% offline!


🌟 Overview

Visually impaired individuals often struggle to identify objects and navigate their surroundings independently. While existing solutions exist, they are often:

  • 🌐 Limited to English only
  • ⚠️ Difficult to operate without visual interaction
  • 🚫 Lacking support for Indian languages
  • ❌ Not optimized for simple, accessible controls
  • 🔌 Require internet connectivity
  • 🔒 Privacy concerns with cloud-based processing

SeeForMe bridges this gap by providing:

✅ Real-time object & scene recognition using on-device AI ✅ Audio feedback in multiple Indian languages ✅ Accessible navigation via simple controls (like volume buttons) ✅ A user-friendly mobile app designed with inclusivity at its core ✅ 100% offline operation - no internet required ✅ Complete privacy - all processing happens on your device


🎯 Key Features

  • 🔍 Object Detection & Scene Understanding – AI-powered recognition using YOLO (80+ object classes)
  • 🧠 Spatial Awareness – Understands object positions (left/right/center, near/far) and relationships
  • 🗣️ Multilingual Audio Output – Support for 10 Indian languages + English
  • 🎧 Natural Language Descriptions – Phi-3 LLM generates clear, spoken descriptions
  • 🎛️ Simple Controls – Operate via volume buttons for hands-free accessibility
  • 📱 Mobile-first Design – Optimized for Android devices
  • 🔒 Privacy-First – All AI processing happens on-device, no data leaves your phone
  • ✈️ Works Offline – No internet connection required

🏗️ Architecture

Offline Processing Pipeline

Camera Frame
    ↓
YOLO Object Detection (TensorFlow Lite)
    ↓
Spatial Scene Analysis
    ↓
Phi-3 Mini LLM (Local Inference)
    ↓
Offline Translation (if needed)
    ↓
Text-to-Speech (Multi-language)

All processing happens on your device!


🛠️ Tech Stack

  • Computer Vision: YOLOv5s FP16 (TensorFlow Lite)
  • Natural Language: Phi-3 Mini (Quantized for mobile)
  • Languages: Dart/Flutter for mobile app
  • Text-to-Speech (TTS): Native platform TTS engines
  • ML Framework: TensorFlow Lite
  • Platform: Android (iOS support available via Flutter)
  • Translation: Offline dictionary-based translation

🌐 Supported Languages

  • 🇬🇧 English (en)
  • 🇮🇳 Hindi (hi)
  • 🇮🇳 Gujarati (gu)
  • 🇮🇳 Bengali (bn)
  • 🇮🇳 Telugu (te)
  • 🇮🇳 Tamil (ta)
  • 🇮🇳 Marathi (mr)
  • 🇮🇳 Kannada (kn)
  • 🇮🇳 Malayalam (ml)
  • 🇮🇳 Punjabi (pa)

📦 Installation & Setup

Prerequisites

  • Flutter SDK (latest stable version)
  • Android Studio / Xcode (for mobile development)
  • TensorFlow Lite models (see setup instructions below)

📸 App Screenshots

Home Screen

Object Detection

Scene Description

---

🛠️ Tech Stack


🧑‍🤝‍🧑 Team & Credits

Developed with ❤️ to make technology more inclusive and accessible.


“See the world, your way.”


About

Cross-platform mobile app assisting visually impaired users through real-time object detection and multilingual audio feedback.

Topics

Resources

Stars

3 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages