moham741/ssl4eo-s12-vits16-data2vec

SSL4EO-S12 — ViT-S/16 pre-trained with Data2Vec

Pre-trained backbone from the SSL4EO-S12 project.

Property	Value
SSL method	Data2Vec
Architecture	ViT-S/16
Input	S2-L1C 13 bands
Pre-training epochs	100
Normalisation	clip [0, 1] by dividing 10 000
Checkpoint	B13_vits16_data2vec_0099_ckpt.pth

Load the backbone

import torch
import timm

model = timm.create_model("vit_small_patch16_224", num_classes=0)
state = torch.load("B13_vits16_data2vec_0099_ckpt.pth", map_location="cpu")

# key may be "model", "state_dict", or "teacher" depending on the method
backbone_state = state.get("model", state.get("state_dict", state))
model.load_state_dict(backbone_state, strict=False)
model.eval()

Citation

@article{wang2022ssl4eo,
  title={SSL4EO-S12: A Large-Scale Multi-Modal, Multi-Temporal Dataset
         for Self-Supervised Learning in Earth Observation},
  author={Wang, Yi and Braham, Nassim Ait Ali and Xiong, Zhitong and
           Liu, Chenying and Albrecht, Conrad M and Zhu, Xiao Xiang},
  journal={arXiv preprint arXiv:2211.07044},
  year={2022}
}