Support for various document upload formats

[Padmaapparao]

need to be able to upload "forms" in text or word documents, scanned images, pdf docs, json files, jpeg/png images, mp4 and other video clips, and audio clips for "Q/A, summarization " etc with OPEA RAG.
Pipeline should be able to consume any type of upload and extract the content (chunk...)

[yongfengdu]

There are multiple dataprep components here(Not supported by helm chart deploy yet), does any of this satisfy your requirement?
https://github.com/opea-project/GenAIComps/tree/main/comps/dataprep

[eero-t]

Discussed this recently with Padma.

@yongfengdu DocSum application supports currently PDF, docx, audio and mp4 video formats.

However, while data-prep service may support also images, DocSum app does not currently use data-prep service.

PS. this ticket would be more relevant for the Comps (or Examples) repo where such support is implemented, than for this (k8s integration) one.

[xiguiw]

need to be able to upload "forms" in text or word documents, scanned images, pdf docs, json files, jpeg/png images, mp4 and other video clips, and audio clips for "Q/A, summarization " etc with OPEA RAG. Pipeline should be able to consume any type of upload and extract the content (chunk...)

This does not a feature request for upload data format.
Q/A RAG does not support image/video yet. For example, what is she summarization of an image?
It's a feature request for RAG support image, video.

The embedding, retrieval, ranking and LLM to support it.

It's a huge feature.
Please submit it in GenAIExamples and provide solid business impact to development team.

[eero-t]

This does not a feature request for upload data format. Q/A RAG does not support image/video yet. For example, what is she summarization of an image?

I think Padma was referring to summarization of its OCR'ed text content. @Padmaapparao ?