FIgLib_ustc_labeled
FIgLib_ustc_labeled extends the original FIgLib training split with PASCAL VOC-style XML bounding-box annotations for wildfire smoke object detection.
Datasets
数据资源
会员审批通过后可下载受限数据。公开数据可直接访问。
SKLFS-WildFire Test
A large-scale real-world wildfire smoke test set containing 50,735 images from 3,309 videos, designed for evaluating open-scene early wildfire detection with difficult negative samples.
SKLFS-WildFire1M
SKLFS-WildFire1M, a large-scale forest smoke detection dataset comprising over one million images. All images are authentic captures from industrial scenarios, having undergone rigorous manual filtering, annotation, and desensitization.
SKLFS-WildFire2025 Test
A large-scale and diverse forest smoke test set containing 8,363 video clips and 152,700 annotated frames, reserved for evaluating smoke detection in real-world surveillance scenarios.