ATHIRU — AI Wildlife Intrusion Monitoring System

ATHIRU is an AI-powered wildlife intrusion monitoring system designed to reduce human–wildlife conflict in forest-adjacent communities.

The system continuously monitors CCTV camera feeds, detects wild animals using YOLO-World, captures evidence, generates real-time alerts, stores incidents in a database, and immediately notifies authorities through Telegram.

---

System Workflow

CCTV Camera / Demo Video
            │
            ▼
     YOLO-World Detection
            │
            ▼
 Multi-Object Tracking (YOLO)
            │
            ▼
 First Detection of Animal
            │
 ┌──────────┼───────────┐
 │          │           │
 ▼          ▼           ▼
Screenshot Database  Telegram Alert
 │          │
 ▼          ▼
FastAPI Backend
 │
 ▼
Live Dashboard Updates

---

Features

• AI-powered wildlife detection using YOLO-World
• Multi-object tracking
• Live CCTV monitoring dashboard
• Camera switching support
• Automatic screenshot capture
• Real-time incident logging
• SQLite database storage
• Live dashboard updates without refresh
• Telegram alert notifications
• Threat statistics
• Incident history
• Camera timestamp overlay
• Location tagging for every alert

---

Tech Stack

Backend

• FastAPI
• SQLAlchemy
• SQLite
• Pydantic
• Jinja2

AI

• YOLO-World
• Ultralytics
• OpenCV

Frontend

• HTML
• CSS
• JavaScript
• Bootstrap 5
• Leaflet.js

Notifications

• Telegram Bot API

---

Project Structure

ATHIRU/

main.py
requirements.txt
.env.example
.gitignore

backend/
│
├── database.py
├── models.py
├── schemas.py
├── telegram_service.py

detector/
│
└── detect.py

video/
│
├── process_video.py
└── stream.py

templates/
│
├── base.html
├── login.html
└── dashboard.html

static/
│
├── css/
├── js/
├── screenshots/
└── videos/
    └── wildlife_demo.mp4   (place manually, see Required Files)

yolov8s-world.pt   (place manually, see Required Files)

---

Installation

Requires Python 3.13. The exact package versions in requirements.txt are pinned to what was tested and confirmed working on Python 3.13 - several of them (torch, numpy, opencv-python, etc.) ship version-specific prebuilt wheels, so installing on a different Python version may fail to find a matching version. If you're on a different Python version, either install Python 3.13 for this project, or relax the pins in requirements.txt (e.g. torch>=2.13.0) and let pip resolve compatible versions for your interpreter.

Requires Git. requirements.txt installs CLIP directly from its GitHub repo (YOLO-World needs it to match text class names like "elephant" to what it sees), so Git must be installed and on your PATH.

Create virtual environment

python -m venv venv

Activate

Windows

venv\Scripts\activate

Linux/Mac

source venv/bin/activate

Install packages

pip install -r requirements.txt

---

Required Files

Place these manually.

yolov8s-world.pt

into

project root

and

wildlife_demo.mp4

into

static/videos/

---

Environment Variables

Copy .env.example to .env and fill in your Telegram bot credentials:

TELEGRAM_BOT_TOKEN=xxxxxxxxxxxxxxxxxxxx

TELEGRAM_CHAT_ID=xxxxxxxxxx

---

Running the Project

Start FastAPI

uvicorn main:app

Open

http://127.0.0.1:8000

---

Running Detection

Open another terminal

python -m video.process_video

---

Demo

1. Login to dashboard.
2. Select CAM-02 Backyard.
3. Demo wildlife video starts.
4. YOLO detects the animal.
5. Screenshot is captured.
6. Incident stored in SQLite.
7. Dashboard updates automatically.
8. Telegram alert is sent with screenshot.
9. Threat statistics update instantly.

---

APIs

Dashboard

GET /

---

Dashboard Page

GET /dashboard

---

Health

GET /api/health

---

Alerts

GET /api/alerts

---

Statistics

GET /api/stats

---

Video Stream

GET /api/stream/backyard

---

Database

SQLite stores

• Species
• Confidence
• Latitude
• Longitude
• Timestamp
• Screenshot path

---

Alert Workflow

Animal Detected
        │
        ▼
First Detection?
        │
      Yes
        │
        ▼
Capture Screenshot
        │
        ▼
Store Alert
        │
        ▼
Update Dashboard
        │
        ▼
Send Telegram Notification

---