feature vector (for vision)

A fixed-length numeric summary of an image, built from shape/texture/color/keypoint statistics, used as a classifier's input.

A vision feature vector packs a fixed number of hand-engineered numeric measurements (shape descriptors, color histograms, texture statistics, a bag-of-visual-words histogram) into one vector per image, so a general-purpose classifier (k-NN, SVM, random forest) can be applied to it exactly as it would to any other tabular feature vector.

Examples

A classical vision pipeline might concatenate a shape's Hu moments, a color histogram, and a bag-of-visual-words vector into one feature vector before handing it to an SVM.