2 ms·
That would be interesting. I think it's what the likes of Narrative are trying to do with the Clip device, but they just take a shot at regular intervals rather
by onion2k 12y ago
That would be interesting. I think it's what the likes of Narrative are trying to do with the Clip device, but they just take a shot at regular intervals rather than trying to be clever about working out what's good algorithmically.
It probably wouldn't be that hard to do based on motion rather than vision if you can make a few assumptions, like 'people still relatively still to look at interesting things' and 'when people are facing each other they want to remember that scene'.