basic questions

Follow the full discussion on Reddit.
Hey I’m trying to get an idea on the scope of work for training a mlm on image and audio data from social media videos, short form so 60 seconds or less. We’d want to look at things like scene changes, object recognition, and video and audio quality, and the spoken/text content. Mostly I just want to know what the data collection would like look like, like would you need 1000s of videos or 100,000s? Roles- can one person person do audio, visual, (speech?) analysis, and data prep, and training, or is there an ideal way to set up a team. And determining timelines. Let me know if we can chat!