tang16 / subspace-learning-for-person-re-identification Goto Github PK

This project forked from handsome1ang/subspace-learning-for-person-re-identification

Working on developing novel algorithms for Person Re-identification, which is the task of identifying the same person across camera networks with non-overlapping views, using Subspace Learning (Metric Learning). Primarily working on developing cross-view semantic learning algorithms,jointly learning common basis matrices,the view-specific semantic projection matrices and association functions for person re-id. Trying to generalize the approach for multi-view analysis to seek a set of linear semantic transforms and relationship mapping functions simultaneously. Working on the VIPeR dataset preliminarily.