Browse Teach Curate

Learned Convolution Kernels as Edge and Texture Detectors

a kernel is where convolution stops being arithmetic and becomes a learned feature detector, and where locality and weight sharing enter as hard-coded assumptions

jeremy Video

Convolutional Neural Networks Feature Detectors and Feature Maps Explained Visually

Convolutional Neural Networks (CNNs) operate on the principle that high-level image features can be hierarchically constructed by detecting low-level patterns through linear convolution operations be…