Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add image retrieval from FAISS index #876

Merged
merged 18 commits into from
Mar 6, 2024

Conversation

mrchtr
Copy link
Contributor

@mrchtr mrchtr commented Feb 26, 2024

The component expects a dataset in the form of:

  • A Dask dataset containing id mapping (image id to image url)
  • The faiss index itself

The component will load the dataset from the remote storage into the component, initialise a CLIP model,
and retrieve similar images based on given prompts.

Copy link
Member

@RobbeSneyders RobbeSneyders left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks @mrchtr!

Would be ideal if we let the user search with embeddings or images as well. Similar to the retrieve_laion_by_embedding component. That should be possible now in a single component with the additionalFields support, although maybe not really nice yet without schema support.

We can do it in a separate PR anyway.

@RobbeSneyders
Copy link
Member

Also, we probably want to use a GPU, I did it here for the embedding component:
https://github.com/ml6team/fondant/blob/main/src/fondant/components/embed_images/src/main.py#L43

Copy link
Member

@RobbeSneyders RobbeSneyders left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks @mrchtr!

@RobbeSneyders RobbeSneyders merged commit 6f78d09 into main Mar 6, 2024
8 of 9 checks passed
@RobbeSneyders RobbeSneyders deleted the feature/add-faiss-based-lioan-retrieval branch March 6, 2024 13:05
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants