Indoor Fire Smoke Dataset

This dataset contains 5,000 images of indoor environments with labeled objects for fire and smoke detection. Each image includes bounding box annotations for detecting visible fire and smoke in various indoor scenes such as homes, offices, and buildings.

The dataset is divided into:

• 70% (3,500 images) for training

• 15% (750 images) for validation

• 15% (750 images) for testing

All images are real and captured under different indoor conditions to support object detection model development for fire and smoke recognition in safety and monitoring systems.