This paper presents an interesting review on the application of artificial intelligence AI algorithms (Machine learning ML and deep learning DL in particular) for processing and analyzing geomatics data. The authors considered in their review only the papers published between 2016 and 2021.

Since the authors reviewed on ML and DL, I think that a brief introduction of these tools and especially the difference between them, will help the readers, that are not familiar to work with, to better understand why there is an increasing demand to use AI.

I appreciated the way the authors took to describe the motivations of the work. But, I think that the fist question that we should ask is : Why researchers are increasingly interested to DL.  Is it because of data complexity only? Efficacity or simplicity of these tools to implement?

On what basis you have selected the “fundamental” sources of Geomatic data?

In section 2.2.2, the authors cited the use of InfraRed Thermography IRT. First, please correct Thermography not termography. You cited methods like Mask R-CNN, MLP or others. I was wondering why there is not the YOLO algorithm, it is one of the most used in object detection and segmentation in visual and infrared images. Besides, I would like to draw your attention that other researchers used image fusion to image preprocessing as a data enhancement method by fusing visible and infrared images.  I raised these remarks since you have compared, in Fig 8, the percentage of papers that used geomatic data with AI and you have concluded in line 540 that IRT data is lower than other types of data.

Please provide more accurate description of the improvements to the state-of-the-art knowledge.

I have other general remarks:

- Please choose between American English or British English --> Analysing and analyzing for example

- The paper is not well revised. There are some grammatical and form errors, ex. line 175, 540…  Also; please correct the legend of Fig. 5

The authors present a review of Artificial Intelligence (AI) approaches to

propose a state of the art based on the analysis of which type of data,

methodology and applications geomatics data are used.

Global overview

Firstly, the authors are thanked for their work which is well structured and well explained.

The objectives of the paper are clear and the reading is eased thanks to a good paper organization.

The authors have made an interesting analysis of the selected publications regarding many criteria that enlighten some trends.

As a consequence, their analysis is deeply linked to their selection of papers which seems to represent a tremendous task.

Even though such selection could be discussed and could lead to inconsistent trends, each topic is explained in detail.

However, the paper form needs to be reviewed.

1) Section 1.3 : Maybe the paper organization should come before Section 1.2? (particularly because Section 1.2 is cited in Section 1.3)

2) Figures : make the figures homogeneous to help the reader. Sometimes there is a title inside the figure, sometimes not.

   Moreover, make sure you have your axes labelled and that labels are set accordingly among the different figures (e.g. Figure 6 VS Figure 8: data types are not in the same order, y-axis label on Figure 6 and not on Figure 8), etc.

Specific remarks / Questions

1) An introduction of Machine Learning (ML) and Deep Learning (DL) and their differences would make sense in this paper,

   particularly because they are mentionned together many times.

2) Could you explain how you selected the pertinent papers (l.87)?

3) Also, have you been able to draw a quick history of the methods and data type/size used over the years that lead the community to this point? 

   This could answer the following question, inherent to your paper: Why researchers are using more and more DL?

4) In your research, how did you considered the papers that use the fusion of data and the combination of AI-based approaches?

5) In your conclusion, you make a comparison of the type of data used over the years (l.625-631). Is it based on Figure 3?

   If so, it means that this conclusion is dependant on the paper selection criteria.

   Did you try to compare your result with the number of matches of your queries based on the keyword and year among the different sources of information?