Extraction and Mapping of Cropland Parcels in Typical Regions of Southern China Using Unmanned Aerial Vehicle Multispectral Images and Deep Learning

Round 1

Reviewer 1 Report

Dear Authors,

 Thank you for the opportunity to review your manuscript submitted for consideration in the Drones journal.

The manuscript is a research article titled “Extraction and Mapping of Cropland Parcels in Typical Regions of Southern China Using Unmanned Aerial Vehicle Multispectral Images and Deep Learning”. Findings from this study may be useful in provide a valuable reference for the extraction of cropland parcels in multiple crop growth stages in southern China or regions with similar characteristics. The manuscript is well structured, well written, and contributes to the scientific community involving “UAV” and “Cropland”. However, there are some concerns that need to be addressed before it may be recommended for publication.

My only concern is in the Introduction.

Introduction could be better structured. The third paragraph starts describing the importance of satellites and previous studies and ends with the description about deep learning. The fourth paragraph starts describing the importance of UAV and previous studies and ends with the description about deep learning. My suggestion is to start the paragraph describing the satellite platform and end with the advantages of UAV and in another paragraph describe deep learning algorithms.

Best regards,

Reviewer

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

The study explored optimal method for accurate extraction of cropland parcels with multiple crop growth stages in southern China from the aspects of deep learning model architecture and data Integration perspectives.

I have some comments that I think will help improve the manuscripts.

1)The introduction talks about the detection algorithms and spatial resolution. However, there is limited discussion relating to terrain information and its potential to improve cropland detection.

2) More discussion about the climate and soil are needed for the study area

3) Lines 185-192- What about radiometric correction of UAV data, how were the reflectance data achieved? Spectral bands should be referenced, and wavelengths described

4) Why was slope generated from DSM not DEM?  DSM generally corporates above ground features

5) How was the data fusion carried out? what algorithm was used or were the data just added directly?

6) Section 2.3-What was the rationale for comparing field measured spectral data with uav data? I think authors should highlight the spectral signatures of the land parcel types as generated with their UAV data.

7) More description is required as to how groundthruthing and validation was conducted.

8) Line 267 Never you begin a sentence with “And”

9) Table 3 Why were the spatial resolutions not resampled to specific interval? That will create a more objective assessment of resolution change.

10) section 2.6 how many data were used for training and validation respectively?

11) Maps have no scale bars and north arrow.

Overall, the findings of the study are interesting for publication in drones after major revision.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

Thanks for revising the manuscript. It reads well.

Author Response

Thank you for your hard work in improving this manuscript.