A minimal IoT-based fire detection system that uses a camera to capture live video frames and a YOLO model trained on a RoboFlow fire dataset to identify fire in real time.
This project combines computer vision and IoT to automatically detect fire.
A custom YOLO model was trained using a RoboFlow fire image dataset, and the system continuously monitors the environment through a camera feed.
When a frame is captured:
- It is processed by the model
- The model predicts if fire is present
- If detected, the system triggers an alert/response
- Dataset: RoboFlow Fire Image Dataset
- Model: YOLO (You Only Look Once)
- Task: Fire vs. No Fire classification + bounding box detection
- Training: Custom training on labeled fire images
- Camera captures live frame
- YOLO model processes the frame
- Fire detected → trigger IoT response
- If not detected → continue monitoring
- Python
- YOLO (Ultralytics / YOLOv5 / YOLOv8)
- OpenCV
- RoboFlow dataset