Next Article in Journal
Bacillus velezensis Associated with Organomineral Fertilizer and Reduced Phosphate Doses Improves Soil Microbial—Chemical Properties and Biomass of Sugarcane
Next Article in Special Issue
Postharvest Geometric Characterization of Table Olive Bruising from 3D Digitalization
Previous Article in Journal
Combining Proximal and Remote Sensors in Spatial Prediction of Five Micronutrients and Soil Texture in a Case Study at Farmland Scale in Southeastern Brazil
Previous Article in Special Issue
The Role of Radiation in the Modelling of Crop Evapotranspiration from Open Field to Indoor Crops
 
 
Article
Peer-Review Record

Detection of Planting Systems in Olive Groves Based on Open-Source, High-Resolution Images and Convolutional Neural Networks

Agronomy 2022, 12(11), 2700; https://doi.org/10.3390/agronomy12112700
by Cristina Martínez-Ruedas 1, Samuel Yanes-Luis 2, Juan Manuel Díaz-Cabrera 3, Daniel Gutiérrez-Reina 2, Rafael Linares-Burgos 4 and Isabel Luisa Castillejo-González 5,*
Reviewer 1:
Reviewer 2: Anonymous
Reviewer 3:
Agronomy 2022, 12(11), 2700; https://doi.org/10.3390/agronomy12112700
Submission received: 5 October 2022 / Revised: 22 October 2022 / Accepted: 25 October 2022 / Published: 31 October 2022

Round 1

Reviewer 1 Report

 In this paper, A CNN model is proposed to distinguish among traditional, intensive, and super-intensive olive groves. This is a highly interesting research on olive groves in Andalusia. But there are still some problems need to be pay attention to:

1.      Has similar work been done in other studies and is the method proposed in this study superior to those in other studies? If so, the authors may add validation for clarification.

 

2.      The three regions on the right in Figure 1 lack arrows pointing to the corresponding positions in the domain on the left.

 

3.      Figure 5. is repeated.

 

4.      Column in Fig. 6 overlaid with legend.

 

5.      Figure 9. and Figure10. are incomplete.

 

6.      Please check and adjust the format of references.

 

7.      There is a lack of corresponding reference at [52].

 

8.      Since the stride size is much smaller than the sub-image size (in a ratio of 1:10) when segmenting the crop images, will it cause data duplication among the training, test and validation sets?

 

9.      Please show the hyperparameters for the training phase of the model.

 

10.  Please describe the metrics for evaluation in detail.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors have done a tremendous job in developing a neural network for recognizing olive groves with various intensification patterns. The work is relevant and carries practical significance. At the same time, there are disadvantages that it is desirable to eliminate before submitting the article for publication:

(The authors did not specify line numbering in the article, so I specify pages and paragraphs to navigate through the article)

1) Paragraph 2.1 does not specify the dimensions (diameter) of the crown of olive trees. How many pixels per olive tree at a resolution of 0.5 m? How does this affect the recognition process?

2) Page 5. Figure 3 goes before its first mention in the text. Should be moved after the first paragraph on page 6.

3) Figure 5 is duplicated on pages 6 and 7.

4) Paragraph 3. There is no verification of the neural network operation by additional methods, for example, manual. (During training, the convolutional neural network collects a set of features for each dataset. In the process of work, it can produce a minimum error of 0.9999, while making mistakes. Additional research should be conducted to check the operation of the neural network and exclude false positive results).

5) Have studies been conducted to solve the problem of recognizing olive groves with different intensification patterns using methods other than those proposed by the authors?

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

This paper is the result of a study on deep learning using CNN, which is sufficient to attract the reader's attention. Also, the figures help to understand the content easily.

But the number of figures is too many. Some figures should be moved to supplements, and I also think that some figures can be represented in tables.

In particular, Figure 5 is a duplicate of two identical flow chart, and Figure 9 does not appear to be a finished making.

I wonder what it would be like to increase the resolution of Figures 10 and 12 further.

It is very well expressed in Figures 8-10. I have a suggestion to the author how to combine the three pictures into one.

I think the method of this paper is a study that can be applied to various crops.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Back to TopTop