Analysis of Statistical and Artificial Intelligence Algorithms for Real-Time Speed Estimation Based on Vehicle Detection with YOLO
, Luis Alberto Morales-Rosales 2,*,†, Rafael Imperial-Rojo 1,†, Mario Alberto Roman-Garay 1,†, Gloria Ekaterine Peralta-Peñuñuri 1,† and Mariana Lobato-Báez 3,†Round 1
Reviewer 1 Report
The authors have reported the study on real-time Speed Estimation based on Vehicle Detection with YOLO which provides good accuracy determined from MAE metrics. In general, the main conclusions presented in the paper are supported by the figures and supporting text. However, to meet the journal quality standards, the following comments need to be addressed.
- Abstract: The authors should highlight more about their model rather than traffic setup. Also provided the general applicability of their model to attract general readers.
- The introduction can be improved and extended. The authors should focus on extending the novelty of the current study.
- If this is the uniqueness of the current study, this point needs to be elaborated with existing references (some of the previous works can be referred as “related works” …“ We summarize the contributions as follows…..” The author did not quite understand the novelty of the model rather than data-set acquisition.
- section 3 : “Preliminars”: I think it is best to invest on current model framework and details rather than well-known models.
- Page 10: “Besides, it identifies the vehicles using the neural network YOLO,..”..YOLO version should be mentioned. Not quite understand why authors selected these two models? Did authors perform ablation study to compare with different models?
- What are the baseline models and benchmark results? The authors compared the result any other existing state-of-the art models?
- Did the authors employ any data augmentation methods before training? If so, it should be mentioned. Also, all hyperparameters (learning rate, mini-batch size, number of epochs, optimizer) and model complexity should be detailed.
- The introduction section needs to elaborate with different detection models . The authors should provide different existing state-of-the-art YOLO model see : Sensors 2021, 21(9), 3263; https://doi.org/10.3390/s21093263, Neural, Comput & Applic (2022) https://doi.org/10.1007/s00521-021-06651-x. Hence they should be mentioned.
- Please provide a fair weakness and limitation of the model, and how it can be improved.
- Typographical errors: There are several minor grammatical errors and incorrect sentence structures. Please run this through a spell checker.
Author Response
Please see the attachment
Author Response File: 
Author Response.pdf
Reviewer 2 Report
General comments.
While the paper is written in a good English, care has not been paid to the proper composition of sentences. Several errors of diverse form are present throughout the work, e.g., “K/H” instead of “km/h” in line 83 and others, “vehicles.li2014Video,kurniawan2018Speed, jalalat2016Vehicle” in line 107 that seems a lacking citation in LaTeX. In addition (and not only), what does “Over Roadway-based classification Unlike previous devices” in line 46 mean? I suggest the authors to fully revise the article searching for and correcting all elements that could decrease readability.
Section 3.
It is not clear how the authors are going to use the proposed methods, i.e., which are the input and output of the models. Even if it will be clearer in Section 4, it is obviously important at least to mention them. Additionally, the methods are not clearly explained. For instance, it is difficult to understand the working principle of Ridge Regression because the meaning of the parameters in Equations 1-3 are not entirely mentioned, as ϑ or J(ϑ). This also applies to the Bayesian Ridge Regression, while Figures 3-5 are not even referenced in the text. The authors are invited to thoroughly revise the entire Section.
Section 4.
- Line 191. Why two cell phones are used and not one alone?
- Line 225. It seems that the videos are taken at 30 fps; still, in line 192, the authors say that the cameras are at 60 fps. Please, elaborate on this element.
Section 5.
It seems that the input data for the statistical and machine learning models are those reported in Table 2. If not, please be more specific. If so, these data are extremely common and easy to retrieve, so that the authors’ best model can be practically employed also by other researchers for their purposes. Is it possible to report the analytical formulation of the best-fit linear regression model so that it can be useful for academics/technicians?
Typos (not limited to the followings)
- Line 138: “describe” instead of “describes”.
- Line 221: “hole” instead of “whole”.
- Line 223: “later use” instead of “later be used”.
- Line 324: “y” instead of “and”, probably untranslated.
- Table 3: “Estadistic models” instead of “Statistical models”, as above.
- Table 3: “RL” / “RC” instead of “LR” / “CR”
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The manuscript can be accepted in its current form.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Reviewer 2 Report
The technical issues highlighted in the previous review have been addressed. The paper is now suitable for publication, provided that the newly introduced problems in the English language are solved (e.g., line 38, line 330, and an entire sentence in Spanish in line 188).
Author Response
Please see the attachment
Author Response File: Author Response.pdf
