Clarify model licensing scope and dataset provenance

Separate implementation licenses from pretrained-weight and training-data terms. Preserve existing valid grants and document unresolved distribution permissions. No weights or access settings changed.

LICENSE-NOTICE.md

@@ -0,0 +1,7 @@
+# svm_emo — licensing scope notice
+
+This distribution retains its existing `license: mit` metadata. Py-Feat's original implementation is MIT; the fitted classifier, scaler and PCA parameters also have training-data provenance.
+
+The [2023 paper and supplementary materials](https://doi.org/10.1007/s42761-023-00191-4) identifies **ExpW, CK+ and JAFFE** as training sources for the emotion SVM. JAFFE is training data for this component. [CK+](https://www.jeffcohn.net/Resources/) and the [JAFFE authors' dataset record](https://zenodo.org/records/14974867) specify noncommercial/research conditions. The [ExpW project page](https://mmlab.ie.cuhk.edu.hk/projects/socialrelation/index.html) does not by itself establish a complete permission grant. Check the agreements applicable to the training run and the scope of fitted-parameter redistribution. See the [dataset register](https://github.com/cosanlab/py-feat/blob/main/LICENSES/DATASETS.md).
+
+This notice clarifies scope; it does not revoke existing valid grants or create a new license for third-party material. Dataset and teacher terms do not automatically relicense every trained artifact or inference output. A software license or model-card badge alone does not establish all checkpoint redistribution or commercial-use permissions.

README.md

@@ -6,6 +6,8 @@ pipeline_tag: image-feature-extraction
 
 # svm_emo
 
+> **Licensing scope:** See the [license and provenance notice](LICENSE-NOTICE.md) before relying on this card's license metadata for pretrained-weight redistribution or commercial use. Existing valid grants are preserved.
+
 ## Model Description
 svm_emo combines histogram of oriented gradient feature extraction with a linear support vector machine to predict emotional face expressions from single frame images. 
 
@@ -98,4 +100,11 @@ print(pred)
 
 ```
 
+## License and provenance
+
+This distribution retains its existing `license: mit` metadata. Py-Feat's original implementation is MIT; the fitted classifier, scaler and PCA parameters also have training-data provenance.
+
+The [2023 paper and supplementary materials](https://doi.org/10.1007/s42761-023-00191-4) identifies **ExpW, CK+ and JAFFE** as training sources for the emotion SVM. JAFFE is training data for this component. [CK+](https://www.jeffcohn.net/Resources/) and the [JAFFE authors' dataset record](https://zenodo.org/records/14974867) specify noncommercial/research conditions. The [ExpW project page](https://mmlab.ie.cuhk.edu.hk/projects/socialrelation/index.html) does not by itself establish a complete permission grant. Check the agreements applicable to the training run and the scope of fitted-parameter redistribution. See the [dataset register](https://github.com/cosanlab/py-feat/blob/main/LICENSES/DATASETS.md).
+
+This notice clarifies scope; it does not revoke existing valid grants or create a new license for third-party material. Dataset and teacher terms do not automatically relicense every trained artifact or inference output. A software license or model-card badge alone does not establish all checkpoint redistribution or commercial-use permissions.