[troyo451]

This is great but does anyone know of essentially the same thing BUT it can actually get content visually? ie., it is great to have a transcript but unless the models can 'see' the video they will miss valuable info like if the in-video captions show a github repo link or something

[broalkvam]

There are many models that extracts context from real time video. But they often just rely on identifying simpel object and tasks. Not longer context spanning a video.