visual-layer / visuallayer Goto Github PK

As a user, I want to be able to get the descriptions, stats and exploration link, without downloading the entire dataset at this moment. Follows the torch convention.

I would like to use your tool to investigate data noise in https://huggingface.co/datasets/aharley/rvl_cdip and https://ds4sd.github.io/icdar23-doclaynet/

It is known in the literature already that there is plenty of noise in RVL_CDIP, yet your tool could provide more quantitative insight.

Hi,
Thanks for amazing work. You have many predefined datasets in your library. My usecase is that I have a dataset which is not listed in your repo. Also, I dont want to upload the dataset to your servers as it is of huge size and also privacy concerns.

Can you show me a way how can I analyze the dataset locally?

Hi,
I saw that you support multiple models to get the embeddings.
https://visual-layer.readme.io/docs/using-your-own-model#running-fastdup-with-a-preconfigured--model

Do you have any benchmarks or any reference for papers which compares these models?

Any guide on when to use what type of model. This would help a lot.

• Pandas warnings are raised when calling explore . Could be fixed probably using a .loc or a .copy in the call. Could also be silenced through the warnings module.
• explore only shows duplicate and leakage for pets dataset - possibly only the top of the issue list is accounted for?
• Document this: .explore loads default of 50 images. Specify num_images as argument to see more.
• add .explore(issue='blur')