Next Article in Journal
Quaternion Wavelet Transform and a Feedforward Neural Network-Aided Intelligent Distributed Optical Fiber Sensing System
Next Article in Special Issue
Multi-Domain Feature Alignment for Face Anti-Spoofing
Previous Article in Journal
Industrial Transfer Learning for Multivariate Time Series Segmentation: A Case Study on Hydraulic Pump Testing Cycles
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Sensors
Volume 23
Issue 7
10.3390/s23073638
Review Report
sensors-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
Article
Peer-Review Record

Polarized Object Surface Reconstruction Algorithm Based on RU-GAN Network

Sensors 2023, 23(7), 3638; https://doi.org/10.3390/s23073638
by Xu Yang 1, Cai Cheng 1, Jin Duan 1,*, You-Fei Hao 1, Yong Zhu 2 and Hao Zhang 1
Reviewer 1:
Jinglun Feng
Reviewer 2:
Ying Zhang
Sensors 2023, 23(7), 3638; https://doi.org/10.3390/s23073638
Submission received: 12 February 2023 / Revised: 7 March 2023 / Accepted: 23 March 2023 / Published: 31 March 2023
(This article belongs to the Special Issue Deep Learning for Computer Vision and Image Processing Sensors)

Round 1

Reviewer 1 Report

This paper describes a proposed framework for surface reconstruction of objects from polarized images only, without the need for additional information such as shadows. The problem of normal ambiguity in the 3D reconstruction process is addressed by designing a U-shaped generative adversarial network (RU-Gan) based on jump compensation to fuse the six feasible solutions for surface normals. The use of jump compensation in the encoder and decoder parts of the network, and reconstruction of the content loss function, among other approaches, are proposed to address this issue. 

However, the reviewer believes the motivation in this paper is not very much clear. The reviewer does not understand what is shown in Line 81-82: At present, a better framework for object surface reconstruction from polarized images only has not been available. Moreover, the current challenges in this field need to be elaborated more. 

The novelty in this manuscript also needs to be improved to show the contribution compared to the SOTA.

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

I have the following questions:

1. What is the innovation of the method? From the perspective of network architecture, the method does not integrate the target characteristics organically and fully. The author still needs to explain this aspect.

2. The experimental object seems to be a little simple, and the analysis of complex targets with multiple structures and errors should be clear, and what is the mathematical dependency between the parameterization requirements, methods and parameters of the corresponding network?

3. Single evaluation index: the author should provide multiple evaluation systems for comparison in terms of error and similarity.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors have addressed the issues regarding the reviews.

Reviewer 2 Report

All my concerns have been fixed

Back to TopTop