about videomme caption question #18
Description
Hi there,
First off, thank you for your valuable work on VideoTree—it's been really useful for my project
I noticed the MME captions you provided in another issue, and I have a couple of questions to better understand their implementation:
what is the frame extraction strategy used when generating these captions?
For instance, how many frames are extracted from the video for processing—is it a fixed number, or does it adjust dynamically based on the video's length?
Looking forward to your response.