Integrating Reverse Engineering for Digital Model Reconstruction and Remanufacturing of Mechanical Components: A Systematic Review

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This is a systematic review paper that is informative, well-organized, and of practical value. It successfully outlines the current landscape of applications, technologies, and challenges of Reverse Engineering (RE) in the field of mechanical components. However, there are still parts of the manuscript that require revision and improvement.
1.The number of references included in this study is insufficient for a review paper. Furthermore, nearly half of the cited literature consists of papers published over a decade ago, which limits the recency and cutting-edge nature of this review.
2.The use of an excessively long table, such as Table 4, is not suitable for an academic paper. A more appropriate format for presentation should be adopted.
3.The discussion should integrate more figures from the cited studies to enhance the article's readability, rather than relying predominantly on textual descriptions.
4.The evaluation method presented in Table 2 is inadequate. The focus should shift towards critiquing the effectiveness and advantages of the methods or algorithms used in the cited studies, rather than merely comparing the technologies they employed.
5.The font styles used in the figures and tables are inconsistent and should be thoroughly checked and standardized.
6.The literature coverage may be incomplete. Relying primarily on databases like Google Scholar and Web of Science, while common, may have led to the omission of relevant studies. The search should be expanded to include specialized engineering databases such as Scopus, IEEE Xplore, and ASME Digital Collection.

Author Response

Reviewer # 1

This is a systematic review paper that is informative, well-organized, and of practical value. It successfully outlines the current landscape of applications, technologies, and challenges of Reverse Engineering (RE) in the field of mechanical components. However, there are still parts of the manuscript that require revision and improvement.

1. The number of references included in this study is insufficient for a review paper. Furthermore, nearly half of the cited literature consists of papers published over a decade ago, which limits the recency and cutting-edge nature of this review.

Response: Thank you for your valuable suggestion. We have included more studies in our manuscript now. Please check the revised manuscript that includes more information from the newly cited studies. Please check pages 13-21 and 31 in the revised manuscript.

//

Newly added references:

Ali, S. A., Khan, M. S., & Stricker, D. (2024, April). BRep Boundary and Junction Detection for CAD Reverse Engineering. In 2024 IEEE 3rd International Conference on Computing and Machine Intelligence (ICMI) (pp. 1-6). IEEE.

Baroiu, N., Moroșanu, G. A., Teodor, V. G., Crăciun, R. S., & Păunoiu, V. (2022). USE OF REVERSE ENGINEERING TECHNIQUES FOR INSPECTING SCREWS SURFACES OF A HELICAL HYDRAULIC PUMP. International Journal of Modern Manufacturing Technologies (IJMMT), 14(2).

Chaudhary, K., & Govil, A. (2021, July). Application of 3D scanning for reverse manufacturing and inspection of mechanical components. In Proceedings of the International Conference on Industrial and Manufacturing Systems (CIMS-2020) Optimization in Industrial and Manufacturing Systems and Applications (pp. 61-76). Cham: Springer International Publishing.

Gabštur, P., Kočiško, M., Kaščak, J., & Pollák, M. (2025). Methodology for Verification of Geometrically Complex Components Through Reverse Engineering. Applied Sciences, 15(7), 3963.

Gálvez, A., Iglesias, A., & Fister, I. (2023, April). Industrial Artificial Intelligence Approach for Shape Reconstruction in Quality Assessment of Digital Data from Manufactured Workpieces. In 2023 4th International Conference on Industrial Engineering and Artificial Intelligence (IEAI) (pp. 86-93). IEEE.

Huo, J., & Yu, X. (2020). Three-dimensional mechanical parts reconstruction technology based on two-dimensional image. International Journal of Advanced Robotic Systems, 17(2), 1729881420910008.

Ktari, A., & Mansori, M. E. (2025). Towards remanufacturing of failed parts through rapid low-pressure sand-casting (LPSC) process. The International Journal of Advanced Manufacturing Technology, 1-20.

Lee, H., Lee, J., Kim, H., & Mun, D. (2022). Dataset and method for deep learning-based reconstruction of 3D CAD models containing machining features for mechanical parts. Journal of Computational Design and Engineering, 9(1), 114-127.

Li, J., He, X., & Li, J. (2015, June). 2D LiDAR and camera fusion in 3D modeling of indoor environment. In 2015 National Aerospace and Electronics Conference (NAECON) (pp. 379-383). IEEE.

Lin, R., Ji, Y., Ding, W., Wu, T., Zhu, Y., & Jiang, M. (2025). A Survey on Deep Learning in 3D CAD Reconstruction. Applied Sciences (2076-3417), 15(12).

Palka, D. (2020). Use of reverse engineering and additive printing in the reconstruction of gears. Multidisciplinary Aspects of Production Engineering, 3(1), 274-284.

Pashkov, D. M., Belyak, O. A., Guda, A. A., & Kolesnikov, V. I. (2022). Reverse engineering of mechanical and tribological properties of coatings: results of machine learning algorithms. Physical Mesomechanics, 25(4), 296-305.

Patpatiya, P., Chaudhary, K., & Kapoor, V. (2022). Reverse manufacturing and 3D inspection of mechanical fasteners fabricated using photopolymer jetting technology. Mapan, 37(4), 753-763.

Raj, T., Hanim Hashim, F., Baseri Huddin, A., Ibrahim, M. F., & Hussain, A. (2020). A survey on LiDAR scanning mechanisms. Electronics, 9(5), 741.

Reddy, G. S., Satyanarayana, V. V., Kumar, J. J., & Reddy, B. R. (2023, November). Deviation analysis of reverse engineered freeform surface with rapid prototyping. In AIP Conference Proceedings (Vol. 2821, No. 1, p. 050009). AIP Publishing LLC.

Rešetar, M., Valjak, F., Branilović, M. G., Šercer, M., & Bojčetić, N. (2024). An approach for reverse engineering and redesign of additive manufactured spare parts. Proceedings of the Design Society, 4, 703-712.

Samavati, T., & Soryani, M. (2023). Deep learning-based 3D reconstruction: a survey. Artificial Intelligence Review, 56(9), 9175-9219.

Shah, G. A., Polette, A., Pernot, J. P., Giannini, F., & Monti, M. (2022). User-driven computer-assisted reverse engineering of editable CAD assembly models. Journal of Computing and Information Science in Engineering, 22(2), 021014.

Todorov, T. T., Gavrilov, T., Semkov, M., & Zagorski, M. (2025, August). Environmentally Sustainable Machining of Complex Parts Using 3D Scanning and Virtual Models. In IOP Conference Series: Earth and Environmental Science (Vol. 1532, No. 1, p. 012035). IOP Publishing.

Vogt, M., Rips, A., & Emmelmann, C. (2021). Comparison of iPad Pro®’s LiDAR and TrueDepth capabilities with an industrial 3D scanning solution. Technologies, 9(2), 25.

Yanamandra, K., Chen, G. L., Xu, X., Mac, G., & Gupta, N. (2020). Reverse engineering of additive manufactured composite part by toolpath reconstruction using imaging and machine learning. Composites Science and Technology, 198, 108318.

//

 

 

2. The use of an excessively long table, such as Table 4, is not suitable for an academic paper. A more appropriate format for presentation should be adopted.

Response: Thank you for the suggestion. We now have reduced the length and revised Table 4. Please check page 33-36 in the revised manuscript.

//

Table 4. Identified Problems and Solutions in RE of Mechanical Parts.

Category	Challenge / Issue Identified	Solution / Recommendation	Author and year
Product Complexity, Internal Geometry, and Physical Barriers	Lack of knowledge about which information is pertinent vs. superfluous	Define a taxonomy and analyze in a controlled reference frame where all info is assumed pertinent	Harston and Mattson (2010)
	Internal/hidden features and hollow geometries are difficult to capture, especially in complex parts	Use multimodal scanning (CT + structured light/laser); recreate hollows via CAD operations (e.g., extruded cuts)	Geng & Bidanda (2017); Cheng et al. (2024); Morato et al. (2021); Yahaya et al. (2023)
	Assemblies with mixed materials (e.g., steel vs. aluminum) create invisibility in scans	Separate components by density or pixel count; export as separate STL files	Roos et al. (2011)
	Absence of 3D CAD data and technical specs for legacy parts	Use RE combined with additive manufacturing (AM) to reconstruct parts from scan data and system analysis	López & Vila (2021)
	CT scans struggle with dense materials due to scatter/beam hardening	Supplement CT with optical scans	Gameros et al. (2015)
	Freeform and curved geometries (e.g., propeller blades) are hard to measure	Application of photogrammetry-based RE (Agisoft Metashape + CAD)	Faizin et al. (2024)
	Existing hole-repair/interpolation methods fail in high-curvature regions	Developed an outlier-plane based hole repair method	Sun et al. (2023)
Tolerance, Dimensional Accuracy, and Error Propagation	Dimensional errors accumulate across RE–AM workflow	Apply tolerance stacking and process control	Geng & Bidanda (2017); Forslund et al. (2018)
	CAD models of worn parts misrepresent original tolerances	Avoid worn regions; apply correction ratios; Use unit step integral iteration method to register damaged point cloud	Jamshidi et al. (2006); Huang et al. (2020)
	Automated tolerance estimation is limited; manual assignment is error-prone and time-consuming	Develop surface texture–based conversion tables; apply hybrid/manual methods (MATLAB, Excel)	Jamshidi et al. (2006); Kaisarlis et al. (2007)
	Non-uniform rational B-spline (NURBS) method captures roughness well but fails to extend internal geometry	Propose a hybrid method using primitives for internal structure and NURBS for external geometry	Helle & Lemu (2021)
	Difficulty in aligning partial scans, defining datums, and managing symmetries	Use Iterative Closest Point (ICP)/global registration; apply knowledge-based rules for datum selection; validate through iterative optimization	Buonamici et al. (2017); Kaisarlis et al. (2007); Freddi et al. (2023)
	Difficulty in assessing measurement uncertainty for internal structures and freeform surfaces	Proposed a modular freeform gage (MFG) using ISO 15530-3 to enable uncertainty estimation and traceability for RE of complex surfaces	Gameros et al. (2015)
Scanning Limitations and Data Acquisition Challenges	Varying and unknown product complexity	Decompose product into information types (e.g., geometry, material)	Harston & Mattson (2010)
	Scanning accuracy affected by material reflectivity, transparency, surface flaws, and environmental conditions	Use powder/matte coatings; maintain stable conditions	Morato et al. (2021); Cheng et al. (2024); Pang & Fard (2020); Tóth & Živčák (2014)
	Multi-sensor data fusion creates alignment and integration issues	Use registration algorithms and AI-based fusion strategies	Cheng et al. (2024)
	STL-based issues: lack of datum planes, poor scaling, no curvature, missing parametric/semantic data	Adjust scaling/planes before CAD import; rebuild geometry via triangulation; apply feature recognition to restore parametric models	Roos et al. (2011); Morato et al. (2021); Forslund et al. (2018)
	Reflective surfaces and misaligned markers cause scan failure	Use reference geometries (e.g., plastic pyramid) and rotate parts	Helle & Lemu (2021)
	Mesh quality issues: noise, holes, discontinuities, outliers, and free-standing triangles	Use mesh cleaning, smoothing, and repair (e.g., Geomagic fill/ bridge/ Relaxpolygons; noise filters)	Deja et al. (2019); Geng & Bidanda (2017); López & Vila (2021); Šagi et al. (2015); Pang & Fard (2020); Yahaya et al. (2023)
	Photogrammetry heavily depends on photo quality, lighting, and angles	Capture dense, well-distributed images (e.g., 40+) and process with advanced software	Faizin et al. (2024)
Process Integration, Post-processing Challenges	Critical manufacturing details (e.g., heat treatment) often missed, leading to part failure	Capture post-processing info alongside geometry/ material	Curtis et al. (2011)
	RE data (point clouds, meshes, models) poorly integrated into PLM	Standardize formats and annotate metadata for traceability	Forslund et al. (2018)
	Difficulty selecting AM/CM processes and materials for spare part recovery	Use structured RE-AM methodology with criteria (lead time, cost, performance)	López & Vila (2021)
	Topology optimization produces non-manufacturable outputs	Re-model in CAD with fillets and smoothed profiles	Pang & Fard (2020)
Human, Material, and Process-Specific Limitations	Lack of operator skills leads to misinterpretation of design intent	Use experienced multidisciplinary teams and specialized RE software	Curtis et al. (2011); Freddi et al. (2023)
	Large point cloud datasets increase processing cost/time	Apply point cloud simplification, mesh decimation, or DL-based compression	Cheng et al. (2024); Morato et al. (2021)
	Lack of tolerance data makes CAD model creation experience-driven	Develop tolerance approximation methods using surface/machining textures	Jamshidi et al. (2006)
	Used parts exhibit uncertain geometry/damage (wear, corrosion, stress)	Apply finite element analysis to predict life and remanufacturing worthiness	Huang et al. (2020)
	Lack of traceability and accuracy in optical scanning validation	Use tactile CMM and modular freeform gages as reference	Gameros et al. (2015)

 

//

 

3. The discussion should integrate more figures from the cited studies to enhance the article's readability, rather than relying predominantly on textual descriptions.

Response: Thank you for your suggestion. We have included more figures from cited studies. Please check pages 15, 17, 18 in the revised manuscript.

//

The generated high-resolution 3D model of a 260 mm propeller blade is shown in Figure 7. Their study highlighted that different reconstruction methods yield varying levels of accuracy, which enables comparison of measurement results and quantification of deviations.

Figure 7. High-resolution 3D reconstruction of a 260 mm propeller blade using Agisoft Metashape and AI-assisted tools (adapted from Faizin et al., 2024).

//

//

NURBS surface reconstruction from point cloud data can be directly performed when the data has clear boundaries or orderly contours, whereas irregular point clouds require segmentation, local surface fitting, and splicing. The process was successfully implemented using Imageware 13.1, and the workflow is illustrated in Figure 6.

Figure 6: Workflow of NURBS surface reconstruction from point cloud data using Imageware 13.1. (Adapted from Dong et al. (2021))

//

//

Figure 8 illustrated the results of reconstructed CAD model from synthetic point cloud data.

Figure 8: Results of point cloud reconstruction, (a) the input point cloud data and (b) the generated CAD model. (Adapted from Zhang et al. (2024))

//

 

4. The evaluation method presented in Table 2 is inadequate. The focus should shift towards critiquing the effectiveness and advantages of the methods or algorithms used in the cited studies, rather than merely comparing the technologies they employed.

Response: Thank you for the insightful comment. We appreciate your observation regarding Table 2. Our intention in presenting Table 2 was not to critique methods in isolation, but rather to provide a concise comparative overview of the technologies employed across the cited studies. The detailed evaluation and critique of the methods, algorithms, and their effectiveness are presented in Sections 3.1–3.4 (Restoration, Remanufacturing, and Redesign; Modeling Internal Geometry, Simulation, and Analysis; AI-Driven RE Approach; and Trends in 3D Scanning). These subsections critically analyze the advantages, limitations, and applications of the methods in their respective contexts. Moreover, we incorporated more studies in Section 3 now. We hope this satisfy your concern.

 

5. The font styles used in the figures and tables are inconsistent and should be thoroughly checked and standardized.

Response: Thank you for pointing this out. We have thoroughly checked all figures and tables and revised them to ensure consistent font styles and formatting throughout the manuscript.

 

6. The literature coverage may be incomplete. Relying primarily on databases like Google Scholar and Web of Science, while common, may have led to the omission of relevant studies. The search should be expanded to include specialized engineering databases such as Scopus, IEEE Xplore, and ASME Digital Collection.

Response: We thank you for this helpful suggestion. In addition to Google Scholar, Web of Science and University of Oklahoma Library, we have expanded our search through Scopus. This ensured broader coverage and helped us capture additional relevant studies in the fields of reverse engineering and mechanical part reconstruction. The revised methodology section reflects this expanded database search, and newly identified studies have been incorporated into the analysis where appropriate. The new studies are included in the manuscript from all the databases. Please check pages 7 and 9 in the revised manuscript.

 

//

Table 1. Research protocol of research.

Review type	Systematic review
Databases	Google Scholar, Web of Science, Scopus, University of Oklahoma library
Paper Search Strategy	Boolean logic using keyword clusters
Paper selection tactics	Based on Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) framework
Language	English only
Timeline (Year)	2005-2025
Inclusion criteria	·        Strong empirical data, case studies, simulations, or experimental results ·        Peer-reviewed research
Exclusion criteria	·        Insufficiently detailed studies on RE in mechanical components ·        Limited relevance to addressing RQs

 

//

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Please see the attachment. 

Comments for author File: Comments.pdf

Author Response

Reviewer # 2

 

The manuscript presents a systematic review on reverse engineering (RE) and digital model reconstruction for remanufacturing mechanical components. The authors have conducted thorough research and provided a comprehensive synthesis of current practices and challenges. They propose pathways to advance RE in industrial applications, particularly in fostering greater automation, accuracy, and integration within digital manufacturing workflows. The work is both interesting and significant. However, there are some key issues that need to be addressed before the manuscript can be considered for publication.

1. First, the definition of the research subject is not sufficiently precise. The manuscript focuses on reverse engineering primarily in terms of the 3D scanning technique for creating 3D models of real-world objects. While 3D scanning is a critical tool in reverse engineering, the field itself is much broader. Reverse engineering involves the analysis of an existing object (physical or digital) to understand its design, structure, function, or manufacturing process, often with the goal of reproducing or improving it. For example, the first sentence of the introduction states: “Reverse engineering (RE) involves the acquisition of the surface geometry of a physical component and the conversion of this geometry into a complete three-dimensional (3D) digital model, using computer-aided design (CAD) software.” It would be more accurate to say “Reverse engineering (RE) in CAD…” because reverse engineering, in general, refers to the systematic process of analyzing a product, system, or piece of software to understand its design, rather than solely acquiring surface geometry.

Response: We appreciate your insightful comment regarding the definition and scope of reverse engineering. We agree that reverse engineering extends beyond 3D scanning to include the systematic analysis of an object’s design, structure, function, or manufacturing process, often with the purpose of reproducing or improving it. In the revised manuscript, we have clarified that our study specifically focuses on reverse engineering in the context of CAD model reconstruction from 3D scanning data of mechanical parts. Accordingly, the first sentence of the introduction has been revised to read.  Please check page 1 in the revised manuscript.

//

Reverse engineering (RE) in computer-aided design (CAD) involves the acquisition of the surface geometry of a physical component and the conversion of this geometry into a complete three-dimensional (3D) digital model.

//
We have also revised the title of the manuscript a bit to better reflect the context.

Previous title

//

Integrating Reverse Engineering and Digital Model Reconstruction for Remanufacturing Mechanical Components: A Systematic Review

//

New title

//

Integrating Reverse Engineering for Digital Model Reconstruction and Remanufacturing of Mechanical Components: A Systematic Review

//

 

2. Additionally, Table 1 compares traditional manufacturing approaches with reverse engineering, implying that data acquisition necessarily requires scanning. However, reverse engineering includes a range of techniques beyond just scanning. The manuscript should be revised to better reflect the broader scope of reverse engineering.

Response: We thank you for this helpful observation. We agree that reverse engineering is not limited to scanning but rather encompasses a range of data acquisition approaches. To address this, we have revised Figure 1 so that the “Data Acquisition” step now reflects broader techniques, including scanning, photogrammetry, coordinate measuring machines (CMMs), tomography/CT, manual measurement, and archival CAD/drawings. This modification ensures the figure more accurately represents the wider scope of reverse engineering. Please check page 2 in the revised manuscript.

//

Figure 1. Comparison of traditional manufacturing vs RE process.

//

 

3. Second, the manuscript is somewhat lengthy, even for a review paper. A more concise version would improve readability and enhance the paper's impact

Response: Thank you for the insightful comment. We appreciate your concern regarding the manuscript length. We carefully revisited the text to identify possible reductions, but given the breadth of the subject and the need to comprehensively cover methodologies, challenges, and recent developments, further shortening would risk omitting essential content. As a systematic review, the paper’s scope inherently requires detailed discussion to maintain completeness and accuracy. Now, We have, however, ensured that the manuscript is well-structured, with clear subsections, tables, and figures to support readability, and we sincerely hope the revised version is represented more structured way.

 

4. In light of these issues, I recommend that the authors carefully revise the manuscript to clarify the research problem and reduce its length.

Response: Thank you for your suggestion. The research problem or research questions are now modified more to reflect the clarity of the manuscript. Please check pages 4 & 5 in the revised manuscript.

//

To systematically address the research gaps identified in the existing literature, this study formulated the following research questions (RQs) to guide its investigation:

RQ1: How is reverse engineering applied in the restoration, remanufacturing, and redesign of mechanical components across different sectors?

RQ2: What methods are most effective for reconstructing internal geometries and integrating RE with simulation and analysis tools?

RQ3: How are AI and automation advancing RE workflows, particularly in data processing, model reconstruction, and error reduction?

RQ4: What are the major challenges, including tolerance deviations, scanning limitations, and human/material factors, that affect the fidelity and usability of RE outputs?

RQ5: What emerging trends and future research directions can enhance the integration of RE with manufacturing processes, and lifecycle management?

//

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

1) It would be nice to have a technology tree. ‘3.4 3D scanning’ is great, but it can seem a bit disconnected from other areas.

2) How about including lidar and camera use in "3.4 3D scanning"? A comparison is needed. It seems better to separate these two and publish them in separate papers.

3) Please explain the difference with the following reference.

A Survey on Deep Learning in 3D CAD Reconstruction

by Ruiquan Lin, Yunwei Ji, Wanting Ding, Tianxiang Wu, Yaosheng Zhu and Mengxi Jiang *ORCID

School of Advanced Manufacturing, Fuzhou University, Quanzhou 362251, China

Appl. Sci. 2025, 15(12), 6681; https://doi.org/10.3390/app15126681

Author Response

Reviewer # 3

 

1. It would be nice to have a technology tree. ‘3.4 3D scanning’ is great, but it can seem a bit disconnected from other areas.

Response: Thank you for this constructive suggestion. In the revised manuscript, we have added a technology tree (new Figure X) to connect Section 3.4 (Trends in 3D Scanning) with the broader reverse engineering workflow. This figure illustrates how different scanning methods link to processing software and CAD platforms. We believe this addition improves the logical flow of the manuscript and ensures Section 3.4 is more clearly integrated with the rest of the study. Please check page 20 in the revised manuscript.

//

The technology tree highlighting 3D scanning and post-scanning technologies are illustrated in Figure X.

Figure 9: Technology tree linking scanning methods, data processing tools, CAD platforms

//

2. How about including lidar and camera use in "3.4 3D scanning"? A comparison is needed. It seems better to separate these two and publish them in separate papers.

Response: Thank you for the suggestion. We have expanded Section 3.4 to include a discussion of LiDAR- and camera-based scanning methods. Please check page 21 in the revised manuscript.

//

Recent developments have also extended scanning approaches to LiDAR–camera fusion and consumer-grade LiDAR sensors. Li et al. (2015) developed a low-cost prototype that fused 2D LiDAR depth data with RGB images from a consumer digital camera, achieving textured 3D indoor reconstructions through checkerboard-based extrinsic calibration and RANSAC alignment. This system demonstrated that camera–LiDAR integration can provide accurate colorized point clouds at a fraction of the cost of 3D LiDAR systems. Similarly, Vogt et al. (2021) evaluated Apple’s iPad Pro LiDAR and TrueDepth sensors against an industrial scanner, showing that while industrial systems remain more precise, consumer devices can achieve sub-millimeter accuracy sufficient for certain reverse engineering tasks and low-cost mass customization. Broader reviews of LiDAR mechanisms (Raj et al., 2020) indicate that MEMS- and solid-state-based systems are increasingly promising, offering advantages in robustness, scalability, and reduced size/weight compared to traditional opto-mechanical setups.

//

//

Newly added references:

Li, J., He, X., & Li, J. (2015, June). 2D LiDAR and camera fusion in 3D modeling of indoor environment. In 2015 National Aerospace and Electronics Conference (NAECON) (pp. 379-383). IEEE.

Vogt, M., Rips, A., & Emmelmann, C. (2021). Comparison of iPad Pro®’s LiDAR and TrueDepth capabilities with an industrial 3D scanning solution. Technologies, 9(2), 25.

Raj, T., Hanim Hashim, F., Baseri Huddin, A., Ibrahim, M. F., & Hussain, A. (2020). A survey on LiDAR scanning mechanisms. Electronics, 9(5), 741.

//

 

3. Please explain the difference with the following reference.

A Survey on Deep Learning in 3D CAD Reconstruction

by Ruiquan Lin, Yunwei Ji, Wanting Ding, Tianxiang Wu, Yaosheng Zhu and Mengxi Jiang *ORCID

School of Advanced Manufacturing, Fuzhou University, Quanzhou 362251, China

Appl. Sci. 2025, 15(12), 6681; https://doi.org/10.3390/app15126681

Response: We thank the reviewer for highlighting the paper by Lin et al. (2025). We have now included this study in our manuscript. Please check page 16-17 in the revised manuscript.

//

Deep learning has recently been positioned as a transformative enabler for CAD reconstruction, with dedicated surveys highlighting the rapid progress in this area. Lin et al. (2025) reviewed approaches spanning point cloud to CAD conversion, sketch-based modeling, B-rep generation, and sketch synthesis, while also mapping the landscape of datasets and metrics. They identified persistent challenges, particularly the scarcity of annotated CAD data, limited constraint handling, and weak integration with industrial toolchains, also pointed to promising directions such as multimodal fusion, Transformer-based models, and interactive AI-assisted design. This survey situates the emerging body of AI-driven RE research within a broader trajectory toward intelligent CAD and digital twin systems.

//

//

Newly added reference:

Lin, R., Ji, Y., Ding, W., Wu, T., Zhu, Y., & Jiang, M. (2025). A Survey on Deep Learning in 3D CAD Reconstruction. Applied Sciences (2076-3417), 15(12).

//

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The author has responded to and revised the manuscript point-by-point in accordance with my comments. I recommend that the manuscript be accepted.

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript shows an improvement compared to the previous version. I have no further comments regarding the content; however, the overall length of 43 pages remains a concern, although this might be a matter for the editor to decide.