Space-Time Image Velocimetry Based on Improved MobileNetV2

Round 1

Reviewer 1 Report

This paper developed a novel flow measurement algorithm by using an improved lightweight network. The developed method performs better than the conventional approaches on different data sets. In general, the presentation of the work is clear and the motivation is well formulated. My minor comments are:

 

1. In the introduction. some contributions of the paper can be highlighted. For instance, how is the MobileNetV2 improved.

2. The typos in the paper need to be corrected. For instance, the title of Section 2. Line 192: “the corresponding frequency...”. Line 13: “my method”->”our method”

3. Give references to VGG16, ResNet50, etc. in Section 4.4 and in the tables.

Author Response

Dear Reviewer：

We quite appreciate for your favorite consideration and the insightful comments. Now we have revised the manuscript (ID: electronics-2148236) exactly according to your comments, and found these comments are very helpful.  The revisions were addressed point by point below：

1. We have highlighted some contributions of the paper in the introduction. Line 62: “We improve the structure of MobileNetV2 by adding dilated convolution and global attention mechanism.”
2. We have corrected the typos. Line 13: “my method”->”our method”. Line 193: “the corresponding frequency...”->”The corresponding frequency...”Line 316：“My model”->”Our model”. The title of Section 2: “Outline”->“Overview”. If we find any other typos we will correct them immediately.
3. We have given references to VGG16, ResNet50, etc. in Section 4.4 and in the tables.

Author Response File: Author Response.pdf

Reviewer 2 Report

The manuscript is well written and needs to be revised before being considered for the possible publication. Following are my queries / suggestions: 

1) What is the reference for the traditional MOT method? 

2) The symbol 's' used in Table 1 and that in Equation 13 may be confused by the reader. I suggest to use different ones. 

3) How the depth of flow 'd' is measured? 

4) In line 275, what is allowed error for Top 5 accuracy?

5) In many places authors used 'my' method. I suggest to use present method. 

6) Line 312, author mentions 'most accurate'. Instead I suggest to use 'very close to the true values measured by current meter method'. Because most accurate needs high level of accuracy!

7) Is current meter readings are ensemble averaged? How many readings are taken at every point? Is current meter is used for surface velocity measurement?

8)  What does authors mean about partial velocity, partial area and partial discharge (Table 5 & 11)?

9) In Table 7, authors mention mean vertical velocity, How it is obtained? Is it point velocity at 0.6 times the depth of flow? please mention. 

10) In Figures 16 and 19, indicate the site details, otherwise they may get confused by the readers. 

11) How did surface velocity coefficients for the two locations are considered? Please highlight. 

12) What are the limitations of STIV? It is limited to only surface velocities. Is it possible to use for velocity profiles in the vertical? 

Author Response

Dear Reviewer:

We are very grateful to your comments for the manuscript (ID: electronics-2148236). According with your advice, we tried our best to amend the relevant part and made some changes in the manuscript. All of your questions were answered below:

1. The reference for the traditional MOT method is

   [9] Fujita, I.; Notoya, Y.; Tani, K.; Tateguchi, S. Efficient and accurate estimation of water surface velocity in STIV. Environmental Fluid Mechanics 2019, 19, 1363–1378.

2. We  have changed the ‘s’ in table 1 to ‘St’.
3.  Our hydrographic stations have professional gauges to measure depth. The data is provided by the staff of the hydrographic station.
4. According to relevant references, the error of MOT is allowed within 3°. So we said the Top-5 Accuracy is considered as the accuracy of MOT detection within the allowed angular error.
5. We have changed "my" to "present" in many places. 
6. We use “very close to the true values measured by current meter method” in Line 313. 
7. The current meter can measure surface velocity as well as underwater velocity. It measures the average velocity over a period of time. The current meter is used by professionals at the hydrographic station and provides us with data.
8. We divide the cross section of the river into several areas according to the starting distance. The area of each region is called a partial area. The volume of water flowing through the area per second is called partial discharge. The average water velocity in the area is called partial velocity. These data are provided by the hydrographic station.

9. In this paper, the mean vertical velocity is calculated from the surface velocity. It is obtained by multiplying the surface velocity by the surface velocity coefficient. This method was mentioned in Section 2.4.

10. We change the names of Figures 16 and 19.

    Figure 16. STIs and their FFT images obtained at Baoji hydrographic Station.

    Figure 19. STIs and their FFT images obtained at Dali hydrographic Station.

11. The surface velocity coefficients for the two locations are obtained after many years of field measurement. They are given by the hydrographic station staff.

12.  We have python code based on the relevant principles and formulas to calculate the average vertical velocity from the surface velocity.

Once again, thank you very much for your comments and suggestions.

Author Response File: Author Response.pdf