A computer doesn't see a picture like you do. It sees a grid of numbers (pixels).
To recognise things, it slides little filters across the grid to find edges, then shapes,
then whole objects. Draw something and watch it happen!
Draw — click the squares (or pick a shape)
the picture (a grid of pixels)
To a computer, your drawing is just this:
1 = filled, 0 = empty. Turn on “show numbers” to see it.
Now find the edges — slide a filter across
the filter (an edge-finder)
edges the filter found
🤖 In a real AI (a CNN)
Real image AIs slide hundreds of filters, stacked in layers: the first layer finds
edges, the next combines edges into shapes (corners, curves), the next into parts
(an eye, a wheel), and the last says what it is. And it learns the best filters by itself
during training — nobody hand-draws them.