You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I'm Niels and work as ML engineer at Hugging Face. I discovered your work through ECCV (we met at your poster presentation!), and indexed your paper here: https://huggingface.co/papers/2309.04820. The paper page lets people discuss about your paper and lets them find artifacts about it (your dataset for instance) you can also claim the paper as yours which will show up on your public profile at HF.
Would you like to host the datasets on https://huggingface.co/datasets? I see you're using your own servers for it. Hosting on Hugging Face will give you more visibility, and will also allow people to do:
In case your dataset consists of images, labels and corresponding JSONs, the easiest might be to start from here: https://huggingface.co/docs/datasets/image_dataset (based on a folder containing your images organized by labels).
Activity
NielsRogge commentedon Nov 3, 2024
Hi,
In case your dataset consists of images, labels and corresponding JSONs, the easiest might be to start from here: https://huggingface.co/docs/datasets/image_dataset (based on a folder containing your images organized by labels).
Alternatively, this could be useful: https://github.com/huggingface/transformers/tree/main/examples/pytorch/semantic-segmentation#creating-a-datasetdict. Basically you start from local lists, then cast the image column to the
Imagefeature.A third option could be to start from a Pandas dataframe (where the JSON data is stored as a dict for each row), then use:
Let me know if you need any help
mahobley commentedon Nov 4, 2024
Closing as there is now a HF datadet
https://huggingface.co/datasets/mahobley/MCAC
NielsRogge commentedon Nov 4, 2024
Great, the viewer looks really nice!
Feel free to link it in the README of this Github repo :)