Skip to content
Learn Motion
ExploreHow it worksMembership
Log in
Learn Motion

How AI Sees: A Conceptual Guide to Image and Video Understanding

1What a Machine Actually Sees2From Pixels to Patterns: Features3How Deep Networks Learn to See4Beyond Labels: Locating and Describing What Is Seen5Adding Time: Understanding Video6How These Systems Learn and How We Judge Them
How Deep Networks Learn to See

Edges, Then Parts, Then Objects

3 / 5
Follow the stack from left to right. The first layer only sees pixels, so its filters can only pick up the simplest local structure: edges, color boundaries, small gradients. Hand that output to the second layer and it is no longer looking at pixels, it is looking at edges. So it can combine edges into slightly larger motifs, a corner, a short curve, a repeated texture. The third layer sees those motifs and can build parts, an eye, a wheel, a patch of fur. Deeper still, parts combine into whole objects. Notice what nobody did: nobody wrote a recipe for an eye detector. The network discovered that one was useful and assembled it out of the edge detectors below. And because each layer feeds the next, changing one early filter changes everything above it, which is why the stack has to be adjusted as a whole.
0:00 / 0:00

Stack convolution layers and something systematic happens. The first layer sees pixels, so its filters can only respond to the simplest local structure: edges, color boundaries, small gradients. Its output map is handed to the second layer, which therefore sees edges rather than pixels. Filters in the second layer can combine edges into slightly larger motifs — a corner, a short curve, a repeated texture. The third layer sees those motifs and can combine them into parts: an eye, a wheel, a patch of fur. Deeper layers see parts and combine them into whole objects.

This is hierarchical feature learning, and it is the answer to the brittleness that ended the previous chapter. Nobody wrote the recipe for an eye detector. The network discovered that an eye detector is useful because it helps reduce error on the training data, and it built one out of the edge detectors below it. The same architecture that finds edges in layer one finds faces in layer twelve, purely because each layer is allowed to re-describe the layer beneath it.

The layers are not independent. Changing a filter in an early layer changes what every later layer receives, so the whole stack has to be adjusted together. That coupling is why training a deep network is a global problem rather than a sequence of local ones.

Previous3 / 5Next

Learn Motion

Generate a course. Learn it properly.

Operated by Wuhan Daoyin Technology Co., Ltd.

Contact: [email protected]
Privacy PolicyTerms of Service

© 2026 Learn Motion