Mora / Datasets / yimengfuyao/detectiumfire
DetectiumFireFree and public
A Multi-modal Dataset Bridging Vision and Language for Fire Understanding. DetectiumFire is a large-scale, multi-modal dataset designed to advance fire understanding in both traditional computer vision and modern vision-language tasks. It provides high-quality real and synthetic fire data, detailed annotations, and human preference feedback for training and evaluating object detectors, diffusion models, and vision-language models (VLMs).
- Published on
Kaggleyimengfuyao/detectiumfire
- Price
- Free
- License
- Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0)
- Allows
- Non-commercial only
- Size
- 102.9 GB
- Tags
- arts and entertainment
- Downloads
- 67,878
- Last updated
- 2025-10-22
Mora did not check this dataset. "Allows" reads the declared license only.