🧠 Divya Drishti: AI-Powered Assistance for the Visually & Hearing Impaired
Empowering visually and hearing-impaired individuals with real-time obstacle detection and sign language recognition using AI.
🎯 Project Overview
Divya Drishti (Sanskrit for Divine Vision) is a dual-module assistive technology suite designed to enhance independence for visually and hearing-impaired individuals. Powered by YOLOv5 for obstacle detection and TensorFlow/Keras for sign language recognition, it delivers real-time audio feedback for navigation and communication.
🦯 Obstacle Detection: Detects obstacles in real-time and provides voice-guided navigation instructions.
✋ Sign Language Detection: Recognizes A-Z hand signs, constructs sentences, and converts them to speech.
⚡ Accessible & Affordable: Built with open-source tools for widespread adoption.
🧭 Navigation
👁️ Obstacle Detection
✋ Sign Language Detection
⚙️ Setup Instructions
📈 Future Enhancements
🎥 Demo Videos
🤝 Team & Credits
👁️ Obstacle Detection
Real-time AI system that detects obstacles and provides audio navigation guidance like “Move Left,” “Go Right,” or “Path Clear.”
💡 Features
🎥 Live Video Feed: Captures and analyzes video via webcam.
🧠 YOLOv5 Model: Uses pretrained YOLOv5 (s/m variant) for robust object detection.
🔊 Voice Guidance: Offline Text-to-Speech (pyttsx3) for directional instructions.
🔍 Obstacle Positioning: Determines obstacle location (left/center/right).
⚙️ REST API: Upload images for detection results and navigation advice.
🚀 GPU Support: CUDA acceleration for faster processing.
📸 Sample Output
Detection: Person
Direction: Move Right
Voice Output: "Move right to avoid person"
Visual Output:
✅ Bounding boxes around detected objects
🏷️ Object class and confidence labels
➡️ Directional arrows for navigation guidance
🚀 Run Locally
git clone https://github.com/CodeClash-Team-Rocket/Divya-Drishti-Models.git
cd Divya-Drishti-Models/Obstacle\ Detection
pip install -r requirements.txt
python main.py
✋ Sign Language Detection
Real-time hand sign recognition (A-Z) with sentence construction and speech output for seamless communication.
🔍 Features
✋ Live Sign Recognition: Detects A-Z hand gestures using webcam.
📃 Sentence Construction: Builds meaningful sentences from sequential signs.
🔈 Speech Synthesis: Converts sentences to audio using pyttsx3.
🧠 Smart Suggestions: PyEnchant-based spell-checker for accurate word output.
🛑 Emergency SOS: Optional button for critical alerts.
📸 Sample Output
Signs Detected: H-E-L-L-O
Sentence: Hello
Voice Output: "Hello"
🚀 Run Locally
cd Divya-Drishti-Models/Sign\ Language\ Detection
pip install -r requirements.txt
python final_preds.py
⚙️ Setup Instructions
Clone the Repository:
git clone https://github.com/CodeClash-Team-Rocket/Divya-Drishti-Models.git
Install Dependencies:
cd Divya-Drishti-Models
pip install -r requirements.txt
Run Modules:
Obstacle Detection: cd Obstacle\ Detection && python main.py
Sign Language Detection: cd Sign\ Language\ Detection && python final_preds.py
API Server: cd Obstacle\ Detection && uvicorn api:app --reload
Requirements:
Python 3.8+ (Obstacle Detection), Python 3.10 (Sign Language Detection)
Webcam for real-time detection
Optional: CUDA-enabled GPU for faster processing
📈 Future Enhancements
🌍 GPS Integration: Enable outdoor navigation with path planning.
🗣️ Multilingual TTS: Support for regional languages.
📱 Mobile App: Deploy API to Android for portable access.
📏 Depth Estimation: Calculate obstacle distances for precise guidance.
.
## 👁️ Obstacle Detection
> **Real-time AI system that detects obstacles and provides audio navigation guidance like “Move Left,” “Go Right,” or “Path Clear.”**
### 💡 Features
- 🎥 **Live Video Feed**: Captures and analyzes video via webcam.
- 🧠 **YOLOv5 Model**: Uses pretrained YOLOv5 (s/m variant) for robust object detection.
- 🔊 **Voice Guidance**: Offline Text-to-Speech (pyttsx3) for directional instructions.
- 🔍 **Obstacle Positioning**: Determines obstacle location (left/center/right).
- ⚙️ **REST API**: Upload images for detection results and navigation advice.
- 🚀 **GPU Support**: CUDA acceleration for faster processing.
### 📦 Tech Stack
| Component | Technology |
|-----------------|-------------------------------|
| Model | YOLOv5 (PyTorch) |
| Voice Engine | pyttsx3 |
| Backend API | FastAPI, Uvicorn |
| Libraries | OpenCV, Pillow, NumPy, Pandas|
| Deployment | Local / Cloud-ready |
### 📸 Sample Output
```text
Detection: Person
Direction: Move Right
Voice Output: "Move right to avoid person"Visual Output:
- ✅ Bounding boxes around detected objects
- 🏷️ Object class and confidence labels
- ➡️ Directional arrows for navigation guidance
git clone https://github.com/CodeClash-Team-Rocket/Divya-Drishti-Models.git
cd Divya-Drishti-Models/Obstacle\ Detection
pip install -r requirements.txt
python main.pyReal-time hand sign recognition (A-Z) with sentence construction and speech output for seamless communication.
- ✋ Live Sign Recognition: Detects A-Z hand gestures using webcam.
- 📃 Sentence Construction: Builds meaningful sentences from sequential signs.
- 🔈 Speech Synthesis: Converts sentences to audio using pyttsx3.
- 🧠 Smart Suggestions: PyEnchant-based spell-checker for accurate word output.
- 🛑 Emergency SOS: Optional button for critical alerts.
| Component | Technology |
|---|---|
| Model | TensorFlow/Keras (CNN) |
| Hand Detection | CVZone, OpenCV |
| GUI | Tkinter |
| Voice Engine | pyttsx3 |
| Spell Check | PyEnchant |
Signs Detected: H-E-L-L-O
Sentence: Hello
Voice Output: "Hello"
cd Divya-Drishti-Models/Sign\ Language\ Detection
pip install -r requirements.txt
python final_preds.py-
Clone the Repository:
git clone https://github.com/CodeClash-Team-Rocket/Divya-Drishti-Models.git
-
Install Dependencies:
cd Divya-Drishti-Models pip install -r requirements.txt -
Run Modules:
- Obstacle Detection:
cd Obstacle\ Detection && python main.py - Sign Language Detection:
cd Sign\ Language\ Detection && python final_preds.py
- Obstacle Detection:
Requirements:
- Python 3.8+ (Obstacle Detection), Python 3.10 (Sign Language Detection)
- Webcam for real-time detection
- Optional: CUDA-enabled GPU for faster processing
- 🌍 GPS Integration: Enable outdoor navigation with path planning.
- 🗣️ Multilingual TTS: Support for regional languages.
- 📱 Mobile App: Deploy API to Android for portable access.
- 📏 Depth Estimation: Calculate obstacle distances for precise guidance.
- 🤖 Wearable Device: Raspberry Pi-based compact solution for portability.
| Module | Demo Link |
|---|---|
| 🦯 Obstacle Detection | Watch Now |
| ✋ Sign Language Detection | Watch Now |