RS pretrained models in huggingface style
Project description
rshf
Remote sensing pretrained models easy loading using huggingface -- PyTorch (for fast benchmarking)
Installation:
pip install rshf
Example:
from rshf.satmae import SatMAE
model = SatMAE.from_pretrained("MVRL/satmae-vitlarge-fmow-pretrain-800")
input = model.transform(torch.randint(0, 256, (224, 224, 3)).float().numpy(), 224).unsqueeze(0)
print(model.forward_encoder(input, mask_ratio=0.0)[0].shape)
TODO:
- Add transforms for each model
- Add Documentation (https://rshf-docs.readthedocs.io/en/latest/)
- Add initial set of models
Citations
| Model Type | Venue | Citation |
|---|---|---|
| BioCLIP | CVPR'24 | link |
| CLIP | ICML'21 | link |
| CROMA | NeurIPS'23 | link |
| GeoCLAP | BMVC'23 | link |
| GeoCLIP | NeurIPS'23 | link |
| Presto | link | |
| Prithvi | link | |
| RemoteCLIP | TGRS'23 | link |
| RVSA | TGRS'22 | link |
| Sat2Cap | EarthVision'24 | link |
| SatClip | link | |
| SatMAE | NeurIPS'22 | link |
| SatMAE++ | CVPR'24 | link |
| ScaleMAE | ICCV'23 | link |
| SINR | ICML'23 | link |
| StreetCLIP | link |
List of models available here: Link
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
rshf-0.0.15.tar.gz
(2.0 MB
view details)
Built Distribution
rshf-0.0.15-py3-none-any.whl
(2.0 MB
view details)
File details
Details for the file rshf-0.0.15.tar.gz.
File metadata
- Download URL: rshf-0.0.15.tar.gz
- Upload date:
- Size: 2.0 MB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/5.1.1 CPython/3.9.20
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 | f1ecce1f32943ba5483b6b1abdd5f6fff6990cf5b8f59a51be97194d0cab57c1 |
|
| MD5 | dba36e7fbb4d573cf725ff40c7af71fb |
|
| BLAKE2b-256 | 618bf275377e88e769ac02912eb4d8ec74ae98369da3af702fa1d0dedc455ce5 |
File details
Details for the file rshf-0.0.15-py3-none-any.whl.
File metadata
- Download URL: rshf-0.0.15-py3-none-any.whl
- Upload date:
- Size: 2.0 MB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/5.1.1 CPython/3.9.20
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 | c7f0114636cb48928e2d6a85fe52892cecb3236e90f930b6a154ac28e173c381 |
|
| MD5 | 79a50008470e2cc09b2ec37bebf0a404 |
|
| BLAKE2b-256 | 144cc5956766150249db865b26b81d7f15c1502c2598ef96ccc3b8dbc127b941 |
