search

assafelovic / gpt-researcher

LLM based autonomous agent that does online comprehensive research on any given topic
https://gptr.dev
Apache License 2.0
14.25k stars 1.86k forks source link

Add supoprt for GPT to research from images .png .jpg .jpeg among other documents #563

Open hemeratrading opened 4 months ago

hemeratrading commented 4 months ago

Currently supported file formats are: PDF, plain text, CSV, Excel, Markdown, PowerPoint, and Word documents.

.png .jpg .jpeg would be really nice to see here on the list in the upcoming updates!

assafelovic commented 3 months ago

Can you give examples of what use cases would images support? I assume this would need vision in models? @hemeratrading

pkandarpa-cs commented 3 months ago

even the ability to extract images found from the web search, and include them in the final report would be good - will require vision so the model knows what the image contains, and where to use that

assafelovic commented 2 months ago

Adding images to the report is very interesting @pkandarpa-cs ! How about helping with a PR for this? Alternatively we'll add it to our backlog

hemeratrading commented 2 months ago

I am sorry for late replay but i thought something like OCR, similar to text just to read images for local document research.

But yes even what @pkandarpa-cs said would be awesome! Thank you for this great app! @assafelovic