RSP-ViTAEv2-S

ViTAE (Vision Transformer with Attention Enhancement) Small model for remote sensing scene classification (51 classes).

Usage

from transformers import AutoModelForImageClassification
import torch

# Load model
model = AutoModelForImageClassification.from_pretrained(
    "BiliSakura/RSP-ViTAEv2-S",
    trust_remote_code=True
)

# Inference
model.eval()
input_image = torch.randn(1, 3, 224, 224)  # (batch, channels, height, width)

with torch.no_grad():
    outputs = model(pixel_values=input_image)
    logits = outputs["logits"]  # Shape: (1, 51)
    predicted_class = logits.argmax(dim=-1).item()

Model Details

• Architecture: ViTAE Transformer
• Input size: 224×224×3
• Number of classes: 51
• Parameters: ~18.9M

Citation

If you use this model, please cite the original RSP paper:

@ARTICLE{rsp,
  author={Wang, Di and Zhang, Jing and Du, Bo and Xia, Gui-Song and Tao, Dacheng},
  journal={IEEE Transactions on Geoscience and Remote Sensing}, 
  title={An Empirical Study of Remote Sensing Pretraining}, 
  year={2023},
  volume={61},
  number={},
  pages={1-20},
  doi={10.1109/TGRS.2022.3176603}
}

Original Repository: ViTAE-Transformer/RSP