About the dataset
Hi,
thanks for the dataset. I have a question: is this the dataset of the paper --> E2E: Onboard satellite real-time classification of thermal hotspots events on optical raw data? This released dataset appears to contain primarily volcanic events, and does not clearly match the full THRAWS dataset (including fires) described in the paper.
Many thanks!
Yes, it was the dataset of that paper. I invite you to checkout the new release: https://www.sciencedirect.com/science/article/pii/S1569843226000117
Thanks for the clarification!
Just to confirm: is the volcanic-only dataset sufficient to reproduce the classification results reported in the paper (e.g., MCC)?
Additionally, I noticed that events such as Spain, Greece, or Latvia (present in the THRAWS metadata and likely corresponding to wildfire events) are not included in the Train/Val/Test splits. Why were these excluded?
I also saw the new segmentation release --> really cool work!