With the release of DINOv3, we can update our model with this new version:
https://ai.meta.com/research/dinov3/
We can also add images from EDFN and LoTSS-HR, such that we can extend our training sample from 3.000 to 12.000 sources.
After re-training and testing, we will upload the best version of the new model to the SURF data repository, which enhances accessibility.
With the release of DINOv3, we can update our model with this new version:
https://ai.meta.com/research/dinov3/
We can also add images from EDFN and LoTSS-HR, such that we can extend our training sample from 3.000 to 12.000 sources.
After re-training and testing, we will upload the best version of the new model to the SURF data repository, which enhances accessibility.