Upload LICENSE.md

LICENSE.md

@@ -0,0 +1,94 @@
+# Licensing and Third-Party Rights Notice
+
+This repository is a curated academic collection derived from multiple third-party image datasets. It does **not** have one uniform permissive license covering all of its contents.
+
+## Hugging Face license field
+
+The dataset card should use:
+
+```yaml
+license: other
+```
+
+This is intentional. Using `mit`, `apache-2.0`, `cc0-1.0`, `cc-by-4.0`, or another single permissive identifier for the complete repository could incorrectly imply that the uploader has permission to relicense all third-party images.
+
+## Original curation materials
+
+To the extent that the repository maintainer owns copyright in the original documentation, preparation notes, and independently authored curation material in this repository, those contributions are made available under the **Creative Commons Attribution 4.0 International License (CC BY 4.0)**.
+
+This permission does not extend to:
+
+- third-party image files;
+- original labels, annotations, or metadata copied from source datasets;
+- third-party papers, descriptions, trademarks, or logos;
+- any content for which the repository maintainer does not hold the necessary rights.
+
+CC BY 4.0 text: https://creativecommons.org/licenses/by/4.0/
+
+## Third-party dataset content
+
+The image files and source-derived annotations remain governed by the rights and terms associated with their original sources. No additional license to those materials is granted by this repository.
+
+### CIFAR-100
+
+Source: https://www.cs.toronto.edu/~kriz/cifar.html
+
+The official CIFAR page provides the dataset for download and requests citation of the associated technical report. The page does not present an explicit standard redistribution license for the image collection. Users should cite the original work and independently assess whether their intended use is permitted.
+
+### Caltech-256
+
+Source: https://data.caltech.edu/records/nyy15-4j048
+
+The CaltechDATA record identifies the resource as an open dataset, but its displayed license field is blank. This repository therefore does not assign a new license to the Caltech-256 images.
+
+### Google Scraped Image Dataset
+
+Source: https://www.kaggle.com/datasets/duttadebadri/image-classification
+
+This collection contains web-scraped images. Individual images may remain protected by third-party copyright or other rights even when the dataset is publicly downloadable. The repository maintainer does not claim ownership of those images and does not grant permission for commercial reuse.
+
+### Oxford Flowers-102
+
+Official source: https://www.robots.ox.ac.uk/~vgg/data/flowers/102/
+
+Package used in the experiments: https://www.kaggle.com/datasets/nunenuh/pytorch-challange-flower-dataset
+
+The official Oxford page provides downloads and citation information but does not display an explicit standard license for the underlying images. A Kaggle mirror may display a platform-level license label, but that label should not automatically be interpreted as proof that all underlying image rights were transferred to the mirror uploader.
+
+### Stanford Dogs
+
+Official source: http://vision.stanford.edu/aditya86/ImageNetDogs/
+
+Package used in the experiments: https://www.kaggle.com/datasets/jessicali9530/stanford-dogs-dataset
+
+This repository does not grant a general commercial or redistribution license for Stanford Dogs images. Users should follow the original dataset terms and any restrictions associated with the source images.
+
+## Recommended distribution approach
+
+Where redistribution rights are uncertain, the safest approach is to distribute only:
+
+- preparation scripts;
+- deterministic train/validation/test indices;
+- split manifests;
+- class mappings;
+- checksums;
+- duplicate and corruption reports;
+- documentation describing how users can obtain the source datasets themselves.
+
+If image files are publicly redistributed, the uploader is responsible for confirming that the relevant source terms and rights permit redistribution.
+
+## Attribution and citation
+
+Use of the prepared splits should acknowledge:
+
+1. the original dataset authors;
+2. the original source or hosting page;
+3. Herman Dihtenko's thesis and preparation pipeline.
+
+Citation information is provided in `README.md`.
+
+## No warranty
+
+The repository is provided for academic and research purposes on an “as is” basis, without warranties or guarantees of title, non-infringement, fitness for a particular purpose, accuracy, or completeness.
+
+This notice is informational and is not legal advice.