Person Re-Identification is the task of matching images of a person across multiple camera views. Almost all prior approaches address this challenge by attempting to learn the possible transformations that relate the different views of a person from a training corpora. Then, they utilize these transformation patterns for matching a query image to those in a gallery image bank at test time.

The pros and Cons of the paper are as follows: the empirical results are strong, showing consistent advantage of the proposed method over previous art (as far as I can judge – I am not familiar with re-identification literature) the new normalized correlation layer is a sensible architecture for the re-identification task. Each input map is compared alone using normalized correlation to a large array of might-ne-relevant positions, providing reach output (1500 output neurons per location – that's a lot) for further processing. More detailed comments: Page 1: Lines 50-53: I did not understand the argument in these lines. It is stated that due to a large search area and inexact matching object parts in image 1 may be matched to background parts in image 2, and this is stated as a remedy to the partial occlusion problem. However, if this happens without significant penalty, irrelevant matches contribute to the score and may come to dominate it if there are many of them.

We present a new family of linear time algorithms based on sufficient statistics for string comparison with mismatches under the string kernels framework. Our algorithms improve theoretical complexity bounds of existing approaches while scaling well with respect to the sequence alphabet size, the number of allowed mismatches and the size of the dataset. In particular, on large alphabets with loose mismatch constraints our algorithms are several orders of magnitude faster than the existing algorithms for string comparison under the mismatch similarity measure. We evaluate our algorithms on synthetic data and real applications in music genre classification, protein remote homology detection and protein fold prediction. The scalability of the algorithms allows us to consider complex sequence transformations, modeled using longer string features and larger numbers of mismatches, leading to a state-of-the-art performance with significantly reduced running times.

We present a new family of linear time algorithms based on sufficient statistics for string comparison with mismatches under the string kernels framework. Our algorithms improve theoretical complexity bounds of existing approaches while scaling well with respect to the sequence alphabet size, the number of allowed mismatches and the size of the dataset. In particular, on large alphabets with loose mismatch constraints our algorithms are several orders of magnitude faster than the existing algorithms for string comparison under the mismatch similarity measure. We evaluate our algorithms on synthetic data and real applications in music genre classification, protein remote homology detection and protein fold prediction. The scalability of the algorithms allows us to consider complex sequence transformations, modeled using longer string features and larger numbers of mismatches, leading to a state-of-the-art performance with significantly reduced running times.

Person Re-Identification is the task of matching images of a person across multiple camera views. Almost all prior approaches address this challenge by attempting to learn the possible transformations that relate the different views of a person from a training corpora. Then, they utilize these transformation patterns for matching a query image to those in a gallery image bank at test time. Deep learning approaches, such as Convolutional Neural Networks (CNN), simultaneously do both and have shown great promise recently. In this work, we propose two CNN-based architectures for Person Re-Identification.