Closed olumolu closed 1 month ago
I will think about it
I've decided to not try to do this, the models that have image recognition capabilities are way to slow to understand a constant stream of frames, even if it was slowed to 1 frame every 5 seconds, that would still be way too fast
Like gpt4o like interface where it can provide help with Share screen Now some new model are coming up with gpt4 like capabilities.