[data request] Ecoset

• Name of dataset: Ecoset
• URL of dataset: https://www.kietzmannlab.org/ecoset/
• Dataset repository: https://codeocean.com/capsule/9570390/tree/v1
• License of dataset: Creative Commons Attribution-NonCommercial-ShareAlike 2.0 license (cc-by-nc-sa-2.0)
• Short description of dataset and use case(s):

Ecoset is a large (155GB) image recognition dataset, combining images of objects with appropriate labels (one label per image). Ecoset is intended to provide higher ecological validity than similar image recognition datasets such as ImageNet (ILSVRC). Ecoset contains 1.5 million images from 565 basic level categories, chosen to be both (i) frequent in linguistic usage, and (ii) rated by human observers as concrete (e.g. ‘table’ is concrete, ‘romance’ is not). Additionally, ecoset was tested for label validity with a mislabelling error rate < 5%, as well as filtered to exclude NSFW content.

For more information on the dataset, consider reading the original publication.

⚠️ Folks who would also like to see this dataset in tensorflow/datasets, please thumbs-up so the developers can know which requests to prioritize. ⚠️

And if you'd like to contribute the dataset (thank you!), see our guide to adding a dataset.

Start with docs/add_dataset.md, then review the Ecoset dataset page, repository, and original publication linked in the issue. Follow the contribution guide to add Ecoset to tensorflow/datasets; done means the dataset contribution meets that guide’s requirements.