A Fast Accurate Attention-Enhanced ResNet Model for Fiber-Optic Distributed Acoustic Sensor (DAS) Signal Recognition in Complicated Urban Environments
Round 1
Reviewer 1 Report
The authors implement and demonstrate that adding an attention module already proposed for sensing improvements based on different well-known platforms, it is possible to improve the system recognition performance. Their model description is well presented, the methodology and experiments are quite appropriated and, the results are clear; in this way, they demonstrate that their proposal operates in real environments further improving the common acoustic recognition systems. In synthesis, they present an interesting work with results that suggest the modifications of such platforms (or at least the 2DCNN and the ResNet. In particular, based on their best results they propose the faster and more accurate attention-based ResNet model.
In regard to the presentation of the work, I have the following suggestions:
In the Introduction section Remove lines 35 to 42:
“The introduction should briefly place the study in a broad context and highlight why it is important. It should define the purpose of the work and its significance. The current state of the research field should be carefully reviewed and key publications cited. Please highlight controversial and diverging hypotheses when necessary. Finally, briefly mention the main aim of the work and highlight the principal conclusions. As far as possible, please keep the introduction comprehensible to scientists outside your particular field of research. References should be numbered in order of appearance and indicated by a numeral or numerals in square brackets—e.g., [1] or [2,3], or [4–6]. See the end of the document for further details on references.”
Finish the phrase (line 47):
“…which has been extensively applied in urban infrastructure monitoring [3-5] and natural disaster prediction [6], etc.”
Improve phrase (Lines 48,49):
“Researchers have made great efforts to improve its hardware index [7-9] and the detection and identification ability [10-17], and a lot of useful feature extraction and classifier design work [18-27] has been involved”
Line 72: Is it correct “top-5 error and top-1 accuracy” ? Please revise.
Line 78: Improve phrase containing “.And…which cannot be adjusted automatically. And the feature extraction is time-consuming”
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 2 Report
In this paper, the authors demonstrated a recognition method for DAS based on an end-to-end attention-based ResNet model. The effectiveness of different time-frequency features of STFT and MFCC are compared as the input of the network. The experiment was well presented. An average accuracy of 99.014% is obtained. Nevertheless, there are some questions that can be raised related to some aspects of the paper:
1. The parameters of the laser and some detailed information of the DAS system should be given.
2. In the discussion part, the ROC cure could be considered in the performance comparison of generalization capability for different networks.
Author Response
Please see the attachment.
Author Response File: Author Response.docx