Skip to content
All work
Computer Vision

The operation behind CNNs, written by hand and compared with SciPy.

Notebook complete; the demo runs the same convolution and correlation in your browser.

NumPyOpenCVSciPyTypeScript (demo)
Convolution versus correlation with an asymmetric kernel

Overview

A manual 2D convolution and correlation implementation with stride and padding options, applied with edge, blur, and sharpen kernels and checked against scipy.signal.convolve2d.

Interactive demo

Stride, valid or same padding, and convolution versus correlation, as in the notebook. The Sobel and Prewitt presets use the standard kernels.