Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

1. The email address of the corresponding author is incorrect, and there are also cross errors in the emails of the second and third authors. Please verify and correct them.
2. Are the references [11-13] on line 37 accurate? What do the quotation marks before the references signify? The references are extremely disorganized. For instance, references 19, 22, 23, 24, 27-29 are not cited anywhere in the main text.
3. The training dataset for the AI image classification model is severely insufficient. The article claims to have trained an image classification model using the Teachable Machine platform, but only utilized 50 labeled images—a dataset too meager for any machine learning model.
4. Table 1 claims that the smart greenhouse system significantly improves upon traditional systems across multiple metrics, but the article fails to specify the exact definition of the traditional system, experimental conditions, or data sources. Are these comparative data derived from parallel controlled experiments conducted simultaneously, or from literature references or historical data? Without controlled experiments conducted under identical environmental conditions, the purported improvement percentages lack scientific validity.
5. In Table 1, the Growth time (days) appears twice: the first instance shows 68 days for the traditional system and 58 days for the smart greenhouse (15% improvement), while the second instance displays 210 and 245 days (17% improvement). The latter values are clearly unreasonable—lettuce's growth cycle typically ranges from 45 to 75 days, making 210 or 245 days impossible. It is speculated that the latter row might actually refer to "Biomass." Such a basic error severely undermines the credibility of the paper, indicating significant negligence in data organization and proofreading by the authors.

 

Author Response

Dear Reviewer 1,

Thank you very much for your careful reading and feedback—it helped us significantly improve the clarity and consistency of the manuscript. In the revised version, we made fine adjustments to both form and content to make the work more robust and easier to verify.

Specifically:

Metadata and editorial consistency.

We corrected and verified author information (emails/affiliations) and reviewed stylistic details that could lead to ambiguity or formal errors.

Tables, ranges, and internal consistency.

We reviewed values ​​and ranges that could be misinterpreted (e.g., reported times/cycles) and added explanatory notes to tables and their footnotes when they involve assumptions or scenarios.

References and citations.
We reorganized and standardized the references to ensure they are in order, have a consistent format, and correspond correctly with the citations in the text. We also removed confusing elements (e.g., unjustified punctuation marks or quotation marks in the bibliography) and adjusted the punctuation.

Clarification of estimated vs. measured results.
To avoid over-interpretation, we explicitly state where there is direct measurement and where there are estimates (e.g., in water/energy indicators), and we strengthen the limitations section to ensure the scope of the study is completely transparent.

Supplementary material.
We have included the relevant code (box-counting and image recognition/processing script) as appendices to improve reproducibility and facilitate technical review.

Thank you again. We believe that with these adjustments, the manuscript is clearer, more consistent, and aligned with good editorial and reproducibility practices.

Sincerely,
[Authors]

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

In general, the title of the paper is relevant and interesting.

 

The abstract does not include the key features of the work regarding new techniques.

The main references should be cited at the end of Figures 1 and 2.

The main motivations of the paper, as well as the main contributions, are not addressed in the introduction.

The paper primarily describes the main characteristics of an intelligent system, without presenting any new techniques.

The evaluation results are not sufficient to assess the performance of the presented work in comparison with other works.

 

Author Response

Dear Reviewer 2:

Thank you very much for your comments and for highlighting the relevance of the topic and title. Your observations were very valuable because they made us realize that, although the work contained several technical elements, they were not sufficiently "visible" in the article's narrative. In the revised version, we reinforced precisely that: novelty, motivation, contributions, and evaluation.

Specifically:

More technical abstract focused on contributions.

We rewrote the abstract to clearly include the techniques and key features: the visual evaluation pipeline (classification + fractal descriptor by box-counting as a complementary indicator), and we clarify from the outset which indicators are estimated (water/energy) and which are model results (e.g., classification performance).

Motivation and contributions are explicitly stated in the Introduction.

We added a brief section outlining the motivation (problem and gap in the literature) and a concise list of main contributions (2–3 items), so the reader can quickly identify what this work contributes beyond simply describing a system.

Citations in Figures 1 and 2. We included key references at the end of the figure captions, as recommended (adapted/inspired from…), reinforcing the traceability and context of these figures.

“It’s not just a system”: we clearly explained the technique and method.
We restructured the methodology section so that the AI ​​and box-counting parts are presented as a reproducible method (not as an anecdote). We also added the code as an appendix to facilitate technical verification.

Evaluation and comparison with other works.

Acknowledging your point, we strengthened the evaluation section: we clarified the model evaluation protocol and added a qualitative comparison/positioning with relevant literature, honestly stating the scope (proof-of-concept) and avoiding claiming “improvements” as if they were the result of a controlled trial.

In short, we addressed your central observation: the manuscript now directly presents the proposed new technique, its rationale, its contributions, and how it was evaluated within the actual scope of the study. Thank you again: your comments significantly improved the clarity and defense of the work.

Sincerely,
[Authors]

We thank the reviewer for the careful reading and constructive feedback. Below we provide point-by-point responses and indicate the changes introduced in the revised manuscript.

 

Comment	Response	Changes in manuscript
1. In general, the title of the paper is relevant and interesting.	Thank you. No change was required.	No change.
2. The abstract does not include the key features of the work regarding new techniques.	We revised the abstract to explicitly state the methodological novelty: the combined use of a lightweight AI image classifier and box-counting fractal descriptors (Df) for non-destructive crop-state assessment. We also clarified that water/energy indicators are estimation-based contextual comparisons.	Abstract: revised to highlight the AI+fractal pipeline and the estimation-based nature of resource indicators.
3. The main references should be cited at the end of Figures 1 and 2.	We revised the captions to place the key reference at the end of each caption, following the reviewer’s suggestion.	Figure 1 caption: revised to end with the main citation [1]. Figure 2 caption: revised to end with the main citation [4].
4. The main motivations of the paper, as well as the main contributions, are not addressed in the introduction.	We added a motivation paragraph and a concise list of contributions at the end of the Introduction to clarify the problem setting and the specific contributions of the manuscript.	Introduction: added motivation paragraph and explicit contribution list.
5. The paper primarily describes the main characteristics of an intelligent system, without presenting any new techniques.	We strengthened the description of the novelty by clarifying the methodological contribution: a complementary image-analysis workflow coupling AI classification with fractal texture descriptors, and the release of MATLAB scripts in the appendices to support reproducibility. We also added a qualitative comparison table to position the work relative to representative studies cited in the manuscript.	Introduction and Results: clarified novelty; Table 2 added; reproducibility statement strengthened; MATLAB scripts already provided in appendices.
6. The evaluation results are not sufficient to assess the performance of the presented work in comparison with other works.	We expanded the evaluation narrative to clarify the dataset (n=550) and how the accuracy metric is produced by the Teachable Machine internal validation procedure, and we added a qualitative comparison (Table 2) to contrast scope, techniques, and reported evidence with representative literature. We also further emphasized that the current results are proof-of-concept and outlined a concrete future-work plan (replicated controlled trials and direct metering) in the limitations section.	Materials and Methods (Phase 3): added evaluation protocol clarification; Results: added Table 2 + contextual discussion; Future work and limitations: reinforced positioning as proof-of-concept and outlined next steps.

 

Sincerely,

The Authors

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

1. The reference citations in this paper are extremely irregular. The references in line 48 are listed as [11][12][13], while line 49 shows reference [4], but shouldn't the in-text citations be arranged in sequential order? This problem occurs more than once throughout the manuscript; please check the entire text.

2. In the manuscript, the authors expanded the dataset from 50 to 550 images. Although this is an increase from the original dataset, the sample size is still too small for deep learning models. Can the trained model achieve expected generalization performance in other greenhouse environments or with different lettuce varieties?

3. The paper only reports 92% accuracy but does not provide detailed metrics such as confusion matrix, precision, recall, and F1 score.

4. The conventional system indicators in Table 1 are derived from literature baselines. It is recommended that the authors explicitly cite specific literature sources and their experimental conditions (such as climate region, planting season, soil type, etc.) in the manuscript, so that readers can evaluate the reasonableness of the comparison.

5. The paper describes the AI chatbot's function of providing planting advice, but lacks user experience evaluation.

6. Please verify whether the format of the appendices complies with the journal's requirements.

Author Response

Manuscript: Smart greenhouse prototype for romaine lettuce integrating IoT, AI-based image classification, and fractal texture analysis

We thank the reviewer for the constructive comments. Below we provide point-by-point responses. All changes were incorporated into the revised manuscript, and key additions are indicated by section/table/figure numbers.

 

Comment 1:

The bibliographic citations are irregular and appear out of sequence in the manuscript.

Response:

We revised the entire manuscript to ensure that in-text citations follow a consistent numeric style and are introduced sequentially. We also consolidated adjacent citations into a single bracket and ensured ascending order within each citation group.

Changes in manuscript:

All in-text citations were reformatted and renumbered; the References section was reordered accordingly.

 

Comment 2:

The dataset size seems small for deep learning models; can the trained model generalize to other greenhouses or lettuce varieties?

Response:

We agree that generalization must be demonstrated beyond a single pilot deployment. In the revised manuscript, we clarify that this work is a proof-of-concept and that the classifier was trained and evaluated on an expanded dataset (n = 1500 images) with an independent held-out test subset (n = 350). We explicitly discuss that performance may vary under different cameras, lighting conditions, greenhouse structures, and lettuce cultivars, and we position multi-site validation as future work.

Changes in manuscript:

Section 2.3 (dataset definition and split); Section 3.1 (discussion of generalization); Section 4 (limitations/future work).

 

Comment 3:

Only accuracy is reported; please include more detailed metrics such as a confusion matrix, precision, and recall.

Response:

We added a complete evaluation of the image classifier, including a confusion matrix and class-wise precision/recall/F1 scores computed on the independent held-out test subset (n = 350). The revised manuscript reports TP, TN, FP, and FN and derives the corresponding metrics to provide a transparent performance characterization.

Changes in manuscript:

Section 3.1 and Table 3 (confusion matrix and precision/recall/F1).

 

Author Response File: Author Response.pdf

Round 3

Reviewer 1 Report

Comments and Suggestions for Authors

1 there is still a problem with the reference. I am not sure whether it is the author's problem or the format of the journal. Please refer to the format of the journal for correction. For example, the reference in line 42 is [4], while the reference in line 43 is [2]
2 the format of the table needs to be reconfirmed. Table 2 is not a three line table, but Table 3 has no lines
3. The overall format needs to be readjusted. There are spaces in paragraph 4. Modify it according to the format requirements and view the full text

Author Response

23 February 2026

Point-by-Point Response to Reviewers

Manuscript ID: inventions-4026536

Title: Smart Greenhouse Integrated with AI, IoT and Renewable Energies for the Optimization of Romaine Lettuce Cultivation

 

Dear Editor and Reviewers,

Thank you for your constructive feedback. We revised the manuscript accordingly and provide below a point-by-point response. Reviewer comments are reproduced in italics, followed by our responses and the locations of changes in the revised manuscript. All modifications are highlighted in the marked-up version.

Summary of key updates in this resubmission:

• Citation consistency: in-text citations were renumbered to follow the sequential order of first appearance, and the reference list was reordered accordingly.
• AI evaluation reporting: the revised manuscript reports an expanded dataset and an independent held-out test subset, including a confusion matrix and precision/recall/F1 metrics.
• Baseline transparency: the “conventional system” (baseline) is explicitly defined and supported by literature sources; baseline comparisons are described as contextual (not a contemporaneous controlled trial).
• Scope clarification: a formal usability/user-experience study of the chatbot was not conducted and is stated as future work.
• Editorial metadata: corresponding-author marks (*) and the email address rgomezb@ecci.edu.co were aligned with the editorial record.

Responses to Reviewer 1

Comment R1.1

Las citas bibliográficas son irregulares y no están ordenadas secuencialmente en el texto.

Response: We reviewed the complete manuscript and renumbered all in-text citations so that reference numbers follow the sequential order of first appearance. We also reordered and renumbered the reference list accordingly and verified that each listed reference is cited in the text.

Revisions: Locations: Introduction and full manuscript (global citation audit); References section (reordered).

Comment R1.2

El conjunto de datos se amplió, pero aún puede ser pequeño para aprendizaje profundo. ¿Generaliza a otros invernaderos o variedades?

Response: We agree. We explicitly position the AI module as a proof-of-concept and report evaluation on an independent held-out test subset. We also added a dedicated limitations/future-work discussion emphasizing that broader generalization requires additional multi-site data (different greenhouses, cameras/lighting, and cultivars).

Revisions: Locations: AI dataset description and split; Results; Limitations/Future Work.

Comment R1.3

Solo se reporta precisión; faltan matriz de confusión, precisión, recuperación y F1.

Response: We added a confusion matrix and class-wise precision/recall/F1 metrics computed on the independent held-out test subset. These results are now included in the Results section and the corresponding table/figure.

Revisions: Locations: Results (AI evaluation) and related table/figure.

Comment R1.4

Los indicadores del sistema convencional en Tabla 1 se derivan de referencias. Citar explícitamente fuentes y condiciones.

Response: We expanded the baseline definition and explicitly cited the literature sources used to derive the Table 1 baseline values. We also clarified that baseline values are literature-supported contextual references and not the outcome of a contemporaneous controlled trial.

Revisions: Locations: Baseline definition section; Table 1 caption/discussion; Limitations.

Comment R1.5

Se describe el chatbot, pero falta evaluación de experiencia de usuario.

Response: We clarified that no formal usability/user-experience evaluation was conducted in this study. We added this as a limitation and propose a structured usability study (e.g., SUS/UEQ) as future work.

Revisions: Locations: Limitations/Future Work; Conclusions.

Comment R1.6

Verificar formato de apéndices.

Response: We reviewed the appendices to ensure consistent formatting and clarified what materials are provided for reproducibility (scripts and parameters).

Revisions: Locations: Appendices A–B and related text.

Responses to Reviewer 2

Comment R2.1

(If applicable) Please ensure that figure captions include the appropriate source citations (e.g., ‘Adapted from …’).

Response: We verified all figure captions and ensured that the appropriate source citations are included at the end of the captions following journal style.

Revisions: Locations: Figure captions where external sources were used.

We hope that the revised manuscript is now suitable for publication. Thank you again for your time and consideration.

Sincerely,

Luis Alejandro Arias Barragan (on behalf of all authors)

Author Response File: Author Response.pdf