Inspiration
We wanted to use Gemini's capabilities, specifically the large context window. Books were the first thing to come to mind. We wanted to see if Gemini could interpret the important emotions through out the books, and give them a ranking. Since Gemini is multimodal, the next step would be for Gemini to interpret a squiggle drawn by hand, which would then be used for find nearest matches to the emotion arcs of books.
What it does
Gemini first analyzes books from Gutenberg, storing the emotions and rankings in a text file for later use. A drawing is then interpreted and used as the search input to find the nearest books.
How we built it
This was done using Python and Streamlit as a front end.
Challenges we ran into
Gemini had trouble interpreting the drawn squiggle with our prompts, so we had to revert to PIL.
Accomplishments that we're proud of
That it worked from start to finish.
What we learned
Some of Gemini's capabilities.
What's next for Gut-emotions
It would be nice to expand on using a squiggle as a search input. This might be helpful for those that are disabled. Searching in other arenas could benefit from this too, as a picture is worth a thousand words.
Log in or sign up for Devpost to join the conversation.