Latent

Convolution explorer

Slide a 3 × 3 kernel across an image and see the arithmetic behind each output pixel. Try edge detectors, blur, and your own kernels.

IntermediateExplained in Convolutions

Convolution: a small window sliding over an image

Each output pixel is the sum of a 3 × 3 patch of input pixels, each multiplied by the matching kernel weight.

Input: drag to move the window
Output: red positive, blue negative, gray near zero
Window
0.900.900.900.900.900.900.900.900.90
×
Kernel
−101−202−101
→
Products
−0.9000.90−1.8001.80−0.9000.90
Σ
Sum0
Kernel
Image
Window positioncolumn 13, row 13
Output at this pixel0
Sobel x: positive where brightness rises left to right, negative where it falls.

Try this

  • Compare Vertical edges and Horizontal edges on the square.
  • Put the window on a flat area: edge kernels output zero there.
  • Choose Custom and make a kernel that detects diagonal edges.

Try "embedding", "softmax", "overfitting", or "backpropagation".