Update dataset card with robotics task, paper link, and usage
Browse filesHi! I'm Niels from the community science team at Hugging Face.
This PR improves the dataset card for the SCFields release by:
- Adding the `robotics` task category to the metadata.
- Linking the associated research paper: [Semantic-Contact Fields for Category-Level Generalizable Tactile Tool Manipulation](https://huggingface.co/papers/2602.13833).
- Adding a sample usage section with a training command found in the official repository.
- Including the BibTeX citation.
README.md
CHANGED
|
@@ -1,15 +1,21 @@
|
|
| 1 |
---
|
| 2 |
license: cc-by-nc-4.0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
---
|
| 4 |
|
| 5 |
# SCFields Release Artifacts
|
| 6 |
|
| 7 |
This dataset repository contains assets, contact-field datasets, and contact-field checkpoints used by the SCFields release code.
|
| 8 |
|
| 9 |
-
|
| 10 |
|
| 11 |
-
|
| 12 |
-
|
| 13 |
|
| 14 |
## Layout
|
| 15 |
|
|
@@ -29,7 +35,11 @@ Each checkpoint folder contains:
|
|
| 29 |
- `model.ckpt`: PyTorch Lightning checkpoint.
|
| 30 |
- `config.yaml`: the accompanying training configuration.
|
| 31 |
|
| 32 |
-
##
|
|
|
|
|
|
|
|
|
|
|
|
|
| 33 |
|
| 34 |
```bash
|
| 35 |
hf download Kevinskwk/scfields-release \
|
|
@@ -47,3 +57,30 @@ The release code's `scripts/download_assets.sh` maps shortened hosted asset path
|
|
| 47 |
assets/peeler_raw -> assets/peeler
|
| 48 |
assets/peeler_combined -> assets/peelers_combined
|
| 49 |
```
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: cc-by-nc-4.0
|
| 3 |
+
task_categories:
|
| 4 |
+
- robotics
|
| 5 |
+
tags:
|
| 6 |
+
- tactile-sensing
|
| 7 |
+
- contact-fields
|
| 8 |
+
- manipulation
|
| 9 |
---
|
| 10 |
|
| 11 |
# SCFields Release Artifacts
|
| 12 |
|
| 13 |
This dataset repository contains assets, contact-field datasets, and contact-field checkpoints used by the SCFields release code.
|
| 14 |
|
| 15 |
+
**Paper:** [Semantic-Contact Fields for Category-Level Generalizable Tactile Tool Manipulation](https://huggingface.co/papers/2602.13833)
|
| 16 |
|
| 17 |
+
**Project page:** https://kevinskwk.github.io/SCFields
|
| 18 |
+
**Code:** https://github.com/Kevinskwk/SCFields
|
| 19 |
|
| 20 |
## Layout
|
| 21 |
|
|
|
|
| 35 |
- `model.ckpt`: PyTorch Lightning checkpoint.
|
| 36 |
- `config.yaml`: the accompanying training configuration.
|
| 37 |
|
| 38 |
+
## Usage
|
| 39 |
+
|
| 40 |
+
### Download
|
| 41 |
+
|
| 42 |
+
You can use the Hugging Face CLI to download the artifacts:
|
| 43 |
|
| 44 |
```bash
|
| 45 |
hf download Kevinskwk/scfields-release \
|
|
|
|
| 57 |
assets/peeler_raw -> assets/peeler
|
| 58 |
assets/peeler_combined -> assets/peelers_combined
|
| 59 |
```
|
| 60 |
+
|
| 61 |
+
### Training Example
|
| 62 |
+
|
| 63 |
+
To train the SCFields policy using the provided dataset and checkpoints, you can use the following command structure:
|
| 64 |
+
|
| 65 |
+
```bash
|
| 66 |
+
python train.py \
|
| 67 |
+
--config-dir=config/scraping_real \
|
| 68 |
+
--config-name=contact_field_delta_ee.yaml \
|
| 69 |
+
data_root=/path/to/scfields \
|
| 70 |
+
task.dataset.dataset_dir=/path/to/scfields/data/real/real_scraper_corrected_lambda1 \
|
| 71 |
+
task.dataset.contact_field_checkpoint_path=/path/to/scfields/checkpoints/contact_field/tools_real/model.ckpt
|
| 72 |
+
```
|
| 73 |
+
|
| 74 |
+
## Citation
|
| 75 |
+
|
| 76 |
+
```bibtex
|
| 77 |
+
@misc{ma2026semanticcontactfieldscategorylevelgeneralizable,
|
| 78 |
+
title={Semantic-Contact Fields for Category-Level Generalizable Tactile Tool Manipulation},
|
| 79 |
+
author={Kevin Yuchen Ma and Heng Zhang and Weisi Lin and Mike Zheng Shou and Yan Wu},
|
| 80 |
+
year={2026},
|
| 81 |
+
eprint={2602.13833},
|
| 82 |
+
archivePrefix={arXiv},
|
| 83 |
+
primaryClass={cs.RO},
|
| 84 |
+
url={https://arxiv.org/abs/2602.13833},
|
| 85 |
+
}
|
| 86 |
+
```
|