touati-kamel/yolov10n-forest-fire-detection

YOLOv10n Forest Fire Detection Model

This model is trained to detect forest fires and related phenomena including:

• Fire
• Fire-smoke
• Fog
• Factory-smoke

Model Details

• Architecture: YOLOv10n
• Input size: 640x640
• Classes: 6 classes
• Training: Enhanced with class separation techniques and checkpointing

Usage

from ultralytics import YOLO

model = YOLO('touatikamel/yolov10n-forest-fire-detection')
results = model.predict('path/to/image.jpg')

Checkpoints

This repository includes training checkpoints saved every 10 epochs:

• checkpoints/best.pt - Best performing model
• checkpoints/latest.pt - Latest checkpoint
• checkpoints/epoch_N.pt - Checkpoint from epoch N

Training Resume

To resume training from a checkpoint:

# Download checkpoint and resume training
checkpoint_path, metadata = download_checkpoint_from_hf("latest")
model = YOLO(checkpoint_path)
model.train(data="path/to/dataset.yaml", resume=True)