3 ms·
the convnet could learn the "mechanical" things you describe, like focusing on the middle / center of the pictures more, etc
by make3 9y ago
the convnet could learn the "mechanical" things you describe, like focusing on the middle / center of the pictures more, etc
- thanatropism 9y agoNo, no, no. The point is feeding the network slightly different images as they're captured by a small field of vision. The real perceived object is not static. You're going to need a temporal, recurrent type of model. Also: a bigger boat.