Estimating the Workability of Concrete with a Stereovision Camera during Mixing

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This manuscript aims to automate the workability of concrete by means of a Computer Vision technique.

The contribution of this study is not significant enough to be published in a journal yet.

My comments to improve the content of this manuscript are as follows:

1.      The proposed method should be novel in contribution to science and technology. Please provide a detailed discussion about contrasting with the state of the art.

2.      Besides the other performance measures specificity and AUC should also be reported for model evaluation.

3.      The authors proposed some ad hoc engineered texture features. It is well known that the texture features are susceptible to the variations in the lighting environment. Authors should discuss robustness of the system under the uncertainty and invariance characteristics of the features utilized.

4.      The PCA analysis and performance results suggest that either the ML model architecture/hyperparameters are not well-tuned, or the feature-based representation, or both, do not distinctly describe the nature under investigation. Authors should improve the system by conducting hyperparameter tuning as well as proposing distinctive features.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The paper presents a series of experiments aimed at evaluating the efficacy of machine learning algorithms in assessing the workability of concrete. It delivers detailed test outcomes and draws conclusions. Nonetheless, the significance of the methodology and findings appears to be somewhat limited.

The Haralick features and the seven selected classifiers are well-known within the field. The authors tested the methods separately and provide the results. However, the results are unsatisfactory, and they did not propose new methods to boost the performance.

The authors are encouraged to try Ensemble methods or DL methods to achieve better results, and resubmit the paper.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf