🚀 Beast Watch

This project uses Google Gemini AI to identify animals from images and provide emergency guidance if the animal is dangerous.

The project is for wildlife detection. There are many villages where wild animals, such as tigers, leopards and wolves, enter unnoticed, and cause chaos and destruction.

We use the power of computer vision and modern technology to help save these animals and the destructions they cause. This goes both ways, as it can save money loss to the villagers and the lives of these animals.

This will soon be connected to an open source computer vision model which monitors CCTV footage (or even your webcam from your phone or laptop), which then processes the data and sends a warning in real-time.

Further ideas are being considered, such as SMS warnings to people in the area, and wildlife officials, and much more! Stay tuned!

🏗 Architecture

Beast Watch operates in a two-stage pipeline:

Stage 1: Object Detection

• Uses YOLO and Meta's AI models to detect animals in an image.
• Filters out false positives (e.g., humans, domestic animals).
• If a wild animal is detected, it proceeds to Stage 2.

Stage 2: LLM Inference & Decision-Making

• A Google Gemini AI model analyzes the detected animal.
• Extracts species-specific data (e.g., speed, behavior, risk level).
• Generates real-time safety instructions based on the animal type.

Emergency Response System

Once an animal is confirmed, Beast Watch triggers multiple response mechanisms:

1. CCTV-Based Iterative Tracking – Continuously monitors the animal's movement.
2. SMS & Audible Alerts – Sends emergency notifications to nearby individuals.
3. Notifying Authorities – Wildlife officials and law enforcement are alerted.
4. Social Media & Press – Broadcasts information to raise awareness.

This structured approach ensures real-time detection, automated warnings, and proactive safety measures for both humans and wildlife.

📌 Features

• Takes an image and a word input.
• Warns the user if the animal is dangerous.
• Provides danger level (0-10), speed comparison, and additional details.
• Optimized for emergency situations (concise yet informative output).

🛠 Setup Instructions

1️⃣ Install Dependencies

pip install google-generativeai pillow python-dotenv

2️⃣ Set Up API Key

Create a key.env file in the project directory and add:

GOOGLE_API_KEY=your_actual_api_key_here

3️⃣ Run the Pipeline

Open and run the Jupyter Notebook (pipeline.ipynb):

from pipeline import process_input  # Ensure your script is named pipeline.ipynb

text_input = "What animal is this?"
image_path = "path/to/your/image.png"

output = process_input(text_input, image_path)
print(output)

👨🏽‍💻👨🏽‍💻👩🏽‍💻👨🏽‍💻👨🏽‍💻👨🏽‍💻 Our Team

1. Mihir Thakur
2. Satvik Panchal
3. Bina Kumari Rauniyar
4. Avi Savalia
5. Taskeen Jafri
6. Himansh Mudigonda

📜 License

This project is open-source. Feel free to modify and improve it!