Model Weights

[Farbdrucker]

Hi,

I am writing from the State Office for Geoinformation and Surveying of Lower Saxony (LGLN) in Germany. We are currently engaged in projects involving soil segmentation and the segmentation of impervious surfaces. You can see a demonstration of our work in this LinkedIn video.

We have been using the model provided by IGNF for our research and found it highly useful. However, we are particularly interested in the Vision Transformer (ViT) model you have developed for soil segmentation. The methodology and results described in your work are very promising.

Would it be possible to get access to the weights of your ViT model? Having access to these weights would greatly aid our research and applications in geoinformation and environmental monitoring.

Thank you for considering our request. We look forward to your response.

Best regards,

Lukas

[clmarmy]

Hi,

Thank you for contacting us. We appreciate your interest in our work.

Please find here the model weights: https://sftpgo.stdl.ch/web/client/pubshares/2HvZAv4VegLzxmXSPbmUeb/browse. As indicated by the file names, both models, the one trained on 10-cm images only and the one trained with mixed resolutions (10, 20 and 40 cm), are available. Let me know if you're missing any information.

Could you update us when you'll have reached at a certain testing/development point with the weights?

Best regards,
Clotilde