D-Fire: Image Dataset for YOLO-Based Fire and Smoke Detection

D-Fire is an image dataset for fire and smoke detection developed by gaia-solutions-on-demand and updated in December 2025. It provides visual data specifically formatted for YOLO object detection tasks to identify environmental hazards.

Use Cases

• Training YOLO models to generate bounding boxes for fire and smoke labels in image frames
• Benchmarking the recall of smoke detection features in early-stage fire scenarios
• Developing automated surveillance systems that trigger alerts based on fire or smoke class identification

Strengths

• Dual-label categorization covering both fire and smoke signatures
• Explicitly tagged for YOLO compatibility, streamlining model training
• Maintained by gaia-solutions-on-demand with updates as recent as 2025

Limitations

• Total image count and class distribution are not specified
• Lack of geographic metadata limits assessment of environmental bias
• Missing license information creates uncertainty for commercial deployment

Provenance

Source

gaia-solutions-on-demand via GitHub

Freshness

Updated as of December 2025.