Paper Title
An Overview on Person Re-Identification Using Camstyle
Abstract
Person re-identification (re-ID) is a kind of process of showing the problemmof matching peopleaacross
disjointtcamera views duringaa multi-camera system. CamStyle can function as knowledge augmentation approach.
Specificallyywith CycleGAN, labeled training images are often style transferred to each camera and alongside the first
training samples form the augmented training set. This type of methodd increases the range inndata with minimal overfittinggand
also it incurs an amount of noise. To urge obviate with noise label smooth regularization (LSR) is make used.
Importantly, CamStyleeareeoften adapted to the challenginggproblemssofone-viewlearningg and unsupervisedddomain
adaptationin the personnre-idbothhof whichhhaveecritical research anddapplication significance.
Keywords - PersonrRe-identification, CamStyle, One-view Learning,UUnsupervised Domain Adaptation.