Questions about results on Dynamicearthnet dataset

[Jack-bo1220]

Thank you for your open source code and weights.

However, I have a question about DynamicEarthNet.

To my knowledge, the GT of the validation set for DynamicEarthNet are not obtainable and can only be submitted to the online leadboard for evaluation (https://codalab.lisn.upsaclay.fr/competitions/2882#results). The MIOU reported in your paper is as high as 51.29, but the SOTA on the online evaluation list is only 37.8. Could you introduce your evaluation strategy?

Thank you and look forward to your reply.

[Jack-bo1220]

Another question is whether the image you selected is the image of Sentinel 2 or the Planet image（3m, RGBNIR）? What is the band information used?