Early Fire and Smoke (YOLOv8)

A YOLOv8n detector finetuned for early domestic fire and smoke: cooking, electrical, and cigarette scenarios. Part of the Model Garden collection of finetuned computer-vision models.

Classes

fire, smoke

Intended use

Designed to pair with a temporal state machine that escalates through clear, smoke present, early warning (growing smoke), fire confirmed, so it alerts on persistent, growing smoke rather than a single noisy frame.

Training

Base: YOLOv8n
Data: D-Fire + Domestic Fire and Smoke
Runtime: ~70 FPS on Apple Silicon (MPS)

Usage

from ultralytics import YOLO

model = YOLO("fire_smoke_yolov8.pt")
results = model("your_video.mp4", stream=True)
for r in results:
    r.plot()

Limitations

Vision alone is never a life-safety system. This is a research tool, not a substitute for a certified smoke or fire detector.

Downloads last month: -