I wanted to try and see if I could use gemini to generate highlights.
This is what the code does:
Takes transcript of a sports game
Uses a Large Language Model (LLM) to pick the top 20 most important moments
Cuts the video based on those moments and stitches them into a highlight reel.
I know there are lots of things to improve from prompt to handling the responses etc and we could try vision capabilities too..but the aim was to see it can work and I guess it can. Here I have shared just 5 minutes of the highlights. I don't own this or the source video, this is just for educational purposes.
Top comments (0)