Learning More in Vehicle Re-Identification: Joint Local Blur Transformation and Adversarial Network Optimization

Round 1

Reviewer 1 Report

GAN based augmentation for vehicle Re-ID has been around for a long while, e.g.,

Yi Zhou, Ling Shao, "Cross-View GAN based Vehicle Gnereation for Re-identification," BMVC 2017

The authors did not elaboratively review this line of research, even though the performance seems to be superior in mots datasets. I can fully appreciate the intention of not generating more training images for different viewing angles, this paper emphasizes the same viewing perspective with GAN generation of local blurring patches to make the Re-ID more robust to different variations, which apparently works from the performance reports.

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Vehicle ReID attempts to locate the identity of a specific vehicle in a huge network of cameras quickly. Vehicle ReID is an important part of smart cities and is widely used in public security, but it is extremely challenging. This paper proposes a novel data augmentation method to improve the performance, and proposes a deep learning framework mainly consists of a local blur transformation and a transformation adversarial module. The idea of this paper is very intuitive, and the writing of this paper is easy to understand. 

1. There are a few spelling and grammatical mistakes in this paper. Please check lines 198, 216, 228, 299, and so on.

2. The weight of the filter matrix for blur transformation has only four non-zero values, that is, only four adjacent pixels are considered. Based on the tradeoff between algorithm performance and resource overhead cost, can the author consider more adjacent pixels?

3. The author claims that this method can be used as a pre-processing layer in other deep learning systems, broadening its application potential. Except the baseline model shown in Figure 7, can the author combine their proposed method with other existing state-of-the-art Vehicle Re-Identification algorithms? 

4. Overall, the literature review should be improved. Without discussing the previous works, the novelty of the paper cannot be assessed properly. It is suggested that the related papers should be cited. For example, DOIs: 10.1016/j.asoc.2022.108485; 10.1080/15732479.2018.1550519; 10.1016/j.jksuci.2022.02.004; 10.1109/TCSVT.2020.3043026.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The article has been revised according to the review comments and can be published without further modification.