This is a fire detection dataset, containing 260 images with multiple instances of fire and smoke, along with their labels. In this repository it is divided using the method 5 fold validation.
Note, the names were changed to fire1,fire2,fire3,...,fire260 for better understanding. In the original dataset the names were longer and with more information.
Each fold contains:
187 train images
21 validation images
52 test images
Each image has its corresponding label in txt form, with identical name.
Original dataset link: https://universe.roboflow.com/cs406/fire-labeling/dataset/2