Animal Pet Rescue

• 

• 

  Telegram alert

Inspiration

During fire emergencies, pets and animals are often trapped and overlooked because first responders cannot see them through heavy smoke. I wanted to use computer vision and AI to give rescue teams real-time vision and instant alerts to save trapped pets before it is too late.

What it does

Animal Pet Rescue monitors live camera feeds to detect both fire hazards and trapped animals simultaneously. When a fire and an animal are identified in the same frame, the system immediately triggers a high-priority alert and sends an instant Telegram notification featuring a timestamped snapshot and visual bounding boxes showing the animal's exact location.

How we built it

The project is built using Python, OpenCV, and Ultralytics YOLO. I implemented a dual-model computer vision pipeline:

1. Fire & Smoke Detection Model: Scans the visual feed for active fire or smoke.
2. Animal Detection Model: Detects pets such as dogs, cats, and other animals.
3. Telegram Bot API: Automates sending real-time emergency alert messages and image snapshots directly to a smartphone.

Challenges we ran into

• Dual-Model Coordination: Running two YOLO models concurrently required optimizing frame processing speeds so alerts trigger without delay.
• API Integration: Setting up the Telegram Bot to reliably transmit live image snapshots alongside alert text during detection events.

Accomplishments that we're proud of

• Successfully building a fully functional end-to-end computer vision pipeline that detects hazards and triggers automated alerts in real-time.
• Using AI technology for a meaningful social cause to help save animals.

What we learned

• How to train and deploy object detection models using Ultralytics YOLO and OpenCV.
• Managing real-time data flows in Python and integrating third-party messaging APIs for automated alerts.

What's next for Animal Pet Rescue

• Integrating thermal imaging camera support to detect animals through dense smoke.
• Expanding the detection dataset to recognize a wider variety of domestic and wild animals.

Built With

• computervision
• machine-learning
• opencv
• python
• telegram
• ultralytics
• yolo