FireSpot-CNX

Authors: Natthaphol P., Warit P., Kanokvate T., Jessada K.

About

The FireSpot database is developed based on a collaboration between National Electronics and Computer Technology Center (NECTEC) and three local municipalities, including Pa Miang, Nong Yaeng, and Choeng Doi, in Chiang Mai, Thailand. In the current release, it consists of 4,000 images. Half of them contain smoke in the early burning stages, and another half do not. Smoke areas in those images are labeled with bounding boxes. The bounding box values are a quadruple $(x_c,y_c,w,h)$, where $(x_c,y_c)$ is the Cartesian coordinates of the point at the box center, $w$ is the box width, and $h$ is the box height. The bounding box values are normalized to the maximum of $1$, i.e., range from $0$ to $1$, of which the point at $(0,0)$ represents the top-left corner, and the point at $(1,1)$ represents the bottom-right corner of the image.

Number of images	Number of bounding boxes
Category # Images Smoke 2,000 None 2,000	Class # Bounding boxes Smoke 2,000

Note: We provide the yolo2pixel function that converts coordinates in the YOLO format to coordinates in pixels.

---

Examples

Download

• FireSpot-CNX-r1 (Images and labels).

Citation

Please cite the following paper if you use our image database:

Pornpholkullapat, N., Phankrawee, W., Boondet, P., Thein, T.L.L., Siharath, P., Cruz, J.D., Marata, K.T., Tungpimolrut, K. and Karnjana, J., 2023, November. FireSpot: A Database for Smoke Detection in Early-stage Wildfires. In 2023 18th International Joint Symposium on Artificial Intelligence and Natural Language Processing (iSAI-NLP) (pp. 1-6). IEEE.

doi: https://doi.org/10.1109/iSAI-NLP60301.2023.10354807

Sponsor & Project members