Self-Organizing Deep Learning (SO-UNet)—A Novel Framework to Classify Urban and Peri-Urban Forests

Round 1

Reviewer 1 Report

The manuscript presents an important and innovative way to classify forests from free Sentinel-2 imagery. The target of the manuscript are urban and peri-urban forests in Lebanon but I believe the method might work on other forests as well. The most outstanding merit of the research reported here is the robustness of classification results of SO-Unet over different imagery and seasons. This is rare indeed.

The only remark I have is that it would be good to see the images in Figures 9 and 10 in larger size, so as to fully appreciate the spatial detail in the classification, and any, however small, differences in the results of SO-Unet. Overall an excellent piece of empirical and innovative research!

Author Response

Dear Prof.,

Thank you very much for reviewing my article and for the encouragement to further improve my research. Please find my answers to your questions in the attached file.

Bets wishes

Author Response File: Author Response.pdf

Reviewer 2 Report

The article "Self-Organizing Deep Learning (SO-UNet) - a novel framework 2
to classify urban and peri-urban forests" generally is a good article, written well with a good structure. I would suggest few following comments for the improvement before it considered for publication. 

L 13. Add the problem of large training samples requirement by deep learning in the abstract and as indicated in the objectives

L160. On line 161 after word “algorithm” add the following “unsupervised Artificial Neural Network (ANN) 

L 280. Please use only no guidance for better clarity and understanding 

L 286. It is better to use over clustering or segmentation instead of classification for SOMs

L 293. Please use patch size instead of image size as indicated in the experimental results

Fig. 5. please add the fig with good quality dpi and pixel value as still it unreadable. 

Lines 303 and  304 should be equation 6

Lines 308 and  310 should be equation 7

Lines 312 and  315 should be equation 8

Line 318. rewrite sentence “For each study area, four images with different dates are selected” to “ For each study area four images are selected with different dates

Line 333 Rewrite “The rule for network size is based on the rule “ to “ the network size is based on the rule”

Change lines 348 and 350 to equation 9

Also need to add the future research work and message for the stakeholders on the basis of findings. 

Author Response

Dear Prof.,

Thank you very much for reviewing my research paper. Your comments were very valuable as they improved the content of the manuscript. Please accept my sincere gratitude for your help and encouragement. Please find attached the answers to your questions.

Best regards

Author Response File: Author Response.pdf