Comment Hardware not YET efficient enough to scare me.. (Score 1) 469
The hardware is not ready, at least not until they use hardware to build composite mutation-images that show relevant (pixel) changes only. There is no point in trying to parse a single image a second, or -on the opposite side- a video stream.
In my opinion, efficient wearable vision software should ignore lower quality versions of what it already saw, it would make a huge efficiency leap. I believe this architecture ultimately would be a software skeleton for a mental world reconstruction much like humans perceive.