Person re‐identification based on a novel mahalanobis distance feature dominated KISS metric learning

A novel and efficient method for person re‐identification based on mahalabobis distance feature is proposed. This feature is built on the existing features but exhibits extraordinary discriminative power. With an additional lightweight PCA, it can be cascaded to greatly enhance the overall discrimination. Then Keep It Simple and Straightforward (KISS) metric learning is conducted on such enhanced features. The approach has been evaluated against current methods on a benchmark dataset and can reach outstanding performance.