BlindSight: AI Scene Describer

🚀 BlindSight is an AI-powered application that provides real-time scene descriptions for visually impaired users. It captures frames from a live video feed, processes them using Google Gemini Pro Vision, and generates detailed textual descriptions of the surroundings.

---

📌 Features

✅ Real-time scene descriptions using AI 📝
✅ Webcam-based video processing 📷
✅ Optimized API usage to reduce quota consumption ⚡
✅ User-friendly Streamlit interface 🎨

---

🛠 Installation

1️⃣ Clone the Repository

git clone https://github.com/yourusername/BlindSight-AI-Scene-Describer.git
cd BlindSight-AI-Scene-Describer

2️⃣ Create a Virtual Environment

python -m venv myenv
source myenv/bin/activate  # macOS/Linux
myenv\Scripts\activate     # Windows

3️⃣ Install Dependencies

pip install -r requirements.txt

4️⃣ Set Up Google Gemini API Key

Replace your API key inside main.py:

API_KEY = "your-google-api-key"

---

▶️ Usage

Run the Streamlit app:

streamlit run main.py

• Click "Start Video Analysis" to begin.
• Click "Stop" to end the analysis.
• The app captures frames and provides real-time AI-generated descriptions.

---

📂 Project Structure

📁 BlindSight-AI-Scene-Describer
 ┣ 📜 main.py             # Main application code
 ┣ 📜 requirements.txt    # Required dependencies
 ┣ 📜 README.md           # Project documentation
 ┗ 📜 .gitignore          # Git ignore file

---

🔧 Requirements

• Python 3.8+
• OpenCV (cv2)
• Streamlit
• Pillow (PIL)
• Google Generative AI SDK

---

📜 License

This project is licensed under the MIT License.

---

🚀 Contributions & Support

• Feel free to open issues or submit pull requests.
• Star ⭐ this repository if you find it helpful!

---