VSainteuf

Hi @fazelkhaha , thanks! - the heatmap.npy is the ground truth for centerness, computed as described in the paper - the zones.npy is the mapping pixel -> object, you can...

Hi @bjovc, Good to know that the dataset preparation worked. You have to compute the normalisation values yourself, it's just the mean and standard deviation of the pixels across the...

Hi @nazib , thanks I appreciate. So the ZONES files contain the pixel to object mapping, you can have a look to [this issue](https://github.com/VSainteuf/utae-paps/issues/14) for more details. For the other...

Thanks for you PR !

Hi @LL0912, No we did not plan to publish a specific repository for our fusion paper. However everything you need is already published. Indeed, the model implementations we used are...

Hi there, Yes indeed the bands are stored in wavelength order. As per the artefact on index 6, I don't really know what this could be. Probably something that happens...

Hi @oguzhannysr, you can have a look at the documentation of the dataset [here](https://github.com/VSainteuf/utae-paps/blob/main/src/dataset.py). The model has been trained on patches of size 128x128x10xT (we don't use bands B01,B09, and...

Hi @manapshymyr-OB, I pasted here the class names of each label in the 44class label set of the dataset. You can then look for samples of the class you are...

Hi @PedroConrado In the paper, the semantic segmentation experiments use a fixed learning rate of 0.001 for 100 epochs. For the panoptic segmentation experiments, we start with a higher learning...