LARGE SCALE PERSON RE-IDENTIFICATION USING PART BASED DEEP HASHING

Abstract

It is important to perform real time search while Large scale Person re-identification on a large gallery. The conventional methods use to focus on discriminative learning which is probabilistic approach of learning. In this proposed project we attempt to use deep learning while integrating it with hashing which provides a framework to evaluate productivity, precision and reliability of Large scale Person re-identification. We use augmentation for creating artificial training images through different ways of processing or combination of multiple processing such as random rotation, shifts, shear and flips etc. We propose Part based deep Hashing (PDH) in which augmented images are the input of deep learning architecture. All the augmented images have different identity. We are using whole image and use it for training deep hashing architecture. We use a ternion loss function which calculates the hamming distance of the pedestrian image. The hamming distance of the images with same identity is smaller than the one with different identity. In this project we use standard large scale dataset specifically Market-1501 & Market-1501 +500K.