pith. sign in

arxiv: 2606.28826 · v1 · pith:G7SMKFVPnew · submitted 2026-06-27 · 💻 cs.CV

RefGlass-GS: A UAV-Enabled Fusion Framework for Photorealistic, Semantic and Interactive Digitization of Reflective Glass Facades via Gaussian Splatting

Zhenyu Liang , Xiao Zhang , Boyu Wang , Zhaolun Liang , Ang Li , Jeff Chak Fu Chan , Mingzhu Wang , Jack C.P. Cheng This is my paper

Pith reviewed 2026-06-30 10:17 UTC · model grok-4.3

classification 💻 cs.CV
keywords Gaussian Splattingreflective glass facadesUAV viewpoint planningglass panel segmentationphotorealistic renderingdigital twinsemantic digitizationdeferred shading
0
0 comments X

The pith

RefGlass-GS combines UAV viewpoint planning with an enhanced Gaussian Splatting pipeline to achieve photorealistic rendering and instance-level segmentation of reflective glass facades.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper presents RefGlass-GS as an end-to-end framework that uses drone-captured images to build digital models of buildings covered in reflective glass. It tackles geometric distortion and unrealistic reflections through a segmentation step that isolates individual panels, an optimization that plans drone paths to capture changing appearances, and modifications to Gaussian Splatting that add a reflection network plus adjusted shading and regularization. These elements together produce object-based models suitable for interactive use in digital twin systems. A reader would care because reflective surfaces defeat standard reconstruction tools, and the reported metric gains indicate a route to reliable virtual inspections and management of modern architecture.

Core claim

The central claim is that a fusion of maximum a posteriori glass panel segmentation with structural regularities, UAV viewpoint planning that maximizes view-dependent coverage, and an optimized Gaussian Splatting model containing a Reflection MLP, deferred shading function, and two enhanced regularization terms produces superior photorealistic, semantic, and interactive digitization of reflective glass facades, with measured gains of 0.1927 mIoU in segmentation, 13.15 dB PSNR in view synthesis, and 5.08 dB PSNR in rendering over prior methods.

What carries the argument

The optimized Gaussian Splatting framework with a Reflection MLP, novel deferred shading function, and two enhanced regularization terms for modeling high-frequency near-field reflections.

If this is right

  • The segmentation step achieves 0.1927 higher mIoU than state-of-the-art methods and is the only approach that extracts instance-level panels.
  • The UAV viewpoint planning function improves novel view synthesis PSNR by 13.15 dB over commercially used nap-of-the-object paths.
  • The full RefGlass-GS modeling pipeline yields an average 5.08 dB PSNR gain over existing Gaussian Splatting techniques on reflective scenes.
  • The standardized data organization converts the representations into object-based models that support interactive facility management on digital twin platforms.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The viewpoint planning objective could be adapted for other view-dependent phenomena such as specular highlights on vehicles.
  • The object-based output format may enable direct linkage between rendered views and maintenance databases without additional manual labeling.
  • The Reflection MLP component might transfer to modeling other near-field reflective effects in non-building environments.

Load-bearing premise

The segmentation method based on maximum a posteriori estimation with structural regularities remains robust to severe reflection and background interference.

What would settle it

A controlled test on a reflective facade scene with strong background interference where the segmentation step fails to produce accurate individual panel masks would disprove the robustness claim.

Figures

Figures reproduced from arXiv: 2606.28826 by Ang Li, Boyu Wang, Jack C.P. Cheng, Jeff Chak Fu Chan, Mingzhu Wang, Xiao Zhang, Zhaolun Liang, Zhenyu Liang.

Figure 1
Figure 1. Figure 1: Overall architecture of the proposed RefGlass-GS fusion framework 3.1 Glass Panel Segmentation based on Structural Regularities To achieve a generalizable glass panel segmentation approach applicable to both planar and curved glass façades, we first describe in Section 3.1.1 how curved glass façades can be effectively rectified into [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 3
Figure 3. Figure 3: Structural linear feature extraction Although each peak in the structure-line intensity histograms can be interpreted as a potential boundary between adjacent glass panels, the significant interference caused by reflective textures and background noise in reflective glass façade scenarios can lead to substantial errors when directly segmenting panels based on structural line intensities alone. Instead, we … view at source ↗
Figure 4
Figure 4. Figure 4: Repetitive patterns represented by similarity matrices After that, the structural line intensities are used as the prior probability, and the similarity matrices are used as the likelihood function to construct the MAP models 𝐻𝑃𝑜𝑠𝑡𝑒𝑟𝑖𝑜𝑟 and 𝑉𝑃𝑜𝑠𝑡𝑒𝑟𝑖𝑜𝑟 for identifying glass panel corner point at horizontal and vertical directions, as shown in Eqs. (5) and (6). 𝐻𝑃𝑜𝑠𝑡𝑒𝑟𝑖𝑜𝑟(𝑖) = ∑ 𝑁𝑜𝑟𝐻𝑆𝐼𝑀(𝑖,𝑗) × 𝑁𝑜𝑟𝐻𝐿𝐼 (𝑗) 𝐻 𝑗… view at source ↗
Figure 5
Figure 5. Figure 5: Glass panel segmentation results: (a) 2D corner point extraction by MAP estimation; (b) 3D panel [PITH_FULL_IMAGE:figures/full_fig_p011_5.png] view at source ↗
Figure 7
Figure 7. Figure 7: Semantic back-projection for glass panel segmentation on second-flight images Subsequently, we aim to obtain sufficiently accurate normal maps to serve as the basis for the Reflection MLP estimation described in Section 3.3.3. The captured images are processed through oblique photogrammetric reconstruction to generate a corresponding 3D mesh model. However, this mesh model exhibits geometric distortions in… view at source ↗
Figure 9
Figure 9. Figure 9: Cuboidal region replacement for façade surface refinement [PITH_FULL_IMAGE:figures/full_fig_p014_9.png] view at source ↗
Figure 11
Figure 11. Figure 11: Interactive facility management of RefGlass [PITH_FULL_IMAGE:figures/full_fig_p017_11.png] view at source ↗
Figure 13
Figure 13. Figure 13: Qualitative comparison of glass panel segmentation. For EBLNet, GhostingNet, and GlassSemNet, [PITH_FULL_IMAGE:figures/full_fig_p020_13.png] view at source ↗
read the original abstract

Existing digitization of buildings with reflective glass facades suffers from geometric reconstruction distortion, unrealistic view-dependent texture rendering, and difficulties in object-based semantic enhancement. Therefore, we propose RefGlass-GS, a fusion framework that enables end-to-end UAV-based photorealistic, semantic, and interactive digitization of reflective glass facades. The contributions include: (1) proposing an individual glass panel segmentation method based on maximum a posteriori estimation with structural regularities, robust to severe reflection and background interference; (2) formulating a UAV viewpoint planning optimization function that maximizes the coverage of view-dependent appearance for sufficient data capture; (3) developing an optimized Gaussian Splatting framework with a Reflection MLP, a novel deferred shading function, and two enhanced regularization terms for effective modeling of high-frequency near-field reflections; (4) introducing a standardized data organization paradigm for structuring GS-based representations into object-based models, facilitating interactive facility management on digital twin platforms. Experiments on real-world reflective glass facade scenes validate the effectiveness and superiority of the proposed method. Specifically, the glass panel segmentation achieves an improvement of 0.1927 in mIoU over SOTA methods, and only our method enables instance-level panel extraction. The UAV view planning improves novel view synthesis for reflective facades by 13.15 dB in PSNR compared to commercially used nap-of-the-object planning methods. The RefGlass-GS modeling outperforms SOTA Gaussian Splatting approaches for reflective scenes with an average improvement of 5.08 dB in PSNR.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 1 minor

Summary. The manuscript proposes RefGlass-GS, a UAV-enabled fusion framework for end-to-end photorealistic, semantic, and interactive digitization of reflective glass facades via Gaussian Splatting. Contributions include (1) a MAP estimation glass-panel segmentation method with structural regularities claimed to be robust to severe reflections, (2) an optimization function for UAV viewpoint planning to maximize view-dependent appearance coverage, (3) an optimized GS model incorporating a Reflection MLP, novel deferred shading, and two enhanced regularization terms for high-frequency near-field reflections, and (4) a standardized data organization paradigm to structure GS representations into object-based models for interactive digital-twin use. Experiments on real-world scenes report a 0.1927 mIoU gain in segmentation (with only this method enabling instance-level extraction), a 13.15 dB PSNR gain from the view planner versus nap-of-the-object baselines, and a 5.08 dB average PSNR gain versus SOTA GS methods for reflective scenes.

Significance. If the empirical improvements and segmentation robustness hold under rigorous validation, the work would advance digital-twin applications for reflective architectural surfaces by combining UAV planning, instance-level semantics, and reflection-aware GS modeling; the explicit integration of a Reflection MLP with deferred shading and the object-based data paradigm are concrete strengths that could enable downstream interactive facility management.

major comments (1)
  1. [Contribution (1)] Contribution (1) and associated experiments: the central claim that MAP estimation plus structural regularities remains robust to severe reflection and background interference is load-bearing for the instance-level extraction, 0.1927 mIoU gain, and downstream object-based GS modeling, yet the manuscript provides no ablation studies, error analysis, or quantitative results on the hardest interference regimes described in the problem statement, leaving the semantic and interactive claims unsupported.
minor comments (1)
  1. [Abstract] The abstract states concrete metric improvements but does not reference the specific tables, figures, or sections containing dataset descriptions, baseline implementations, or error bars; this should be added for verifiability even if the full experimental section exists.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive critique. The single major comment is addressed point-by-point below. We agree that additional targeted validation is warranted and will revise accordingly.

read point-by-point responses

  1. Referee: [Contribution (1)] Contribution (1) and associated experiments: the central claim that MAP estimation plus structural regularities remains robust to severe reflection and background interference is load-bearing for the instance-level extraction, 0.1927 mIoU gain, and downstream object-based GS modeling, yet the manuscript provides no ablation studies, error analysis, or quantitative results on the hardest interference regimes described in the problem statement, leaving the semantic and interactive claims unsupported.

    Authors: We accept the referee's assessment that the robustness claim under the most severe reflection and background conditions is central yet insufficiently supported by dedicated quantitative analysis. The current manuscript reports aggregate mIoU gains and qualitative instance-level results across real-world scenes, but does not isolate performance on the hardest interference subsets or provide ablations of the structural-regularity priors. In the revised version we will add: (i) an ablation table removing each structural regularity term in turn, (ii) error analysis (failure-case images and per-scene metrics) on the subset of views exhibiting the strongest reflections and background clutter, and (iii) quantitative mIoU and instance-extraction success rates restricted to those hardest regimes. These additions will directly substantiate the load-bearing claim. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical framework with no derivations or self-referential reductions

full rationale

The paper proposes a multi-component framework (MAP segmentation with structural regularities, UAV view planning optimization, Reflection MLP + deferred shading in GS, and object-based data organization) whose claims rest entirely on reported experimental metrics (0.1927 mIoU gain, 13.15 dB and 5.08 dB PSNR improvements). No equations, fitted parameters presented as predictions, self-citation load-bearing uniqueness theorems, or ansatzes smuggled via prior work appear in the provided text. All load-bearing steps are externally falsifiable via the described real-world scene experiments rather than reducing to input definitions by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Abstract-only review provides no explicit free parameters, new physical entities, or detailed axioms beyond domain assumptions about glass panel structure; full paper would be needed to audit these.

axioms (1)
  • domain assumption Glass panels exhibit detectable structural regularities that support MAP estimation despite reflections.
    Invoked in the individual glass panel segmentation method.

pith-pipeline@v0.9.1-grok · 5846 in / 1267 out tokens · 41298 ms · 2026-06-30T10:17:53.288485+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

65 extracted references · 60 canonical work pages · 2 internal anchors

  1. [1]

    Y. Lv, Y. Pang, L. Ju, W. Xu, Z. Liang, J. He, L. Zhang, Automating digitization of engineered slopes to support interactive digital twin visualization, Automation in Construction 181 (2026) 106564. https://doi.org/10.1016/j.autcon.2025.106564

    work page doi:10.1016/j.autcon.2025.106564 2026

  2. [2]

    Mousavi, M

    V. Mousavi, M. Rashidi, G. Shayan, B. Samali, A data -driven digital twin model for bridge health monitoring using feature fusion and unsupervised deep learning, Information Fusion (2025) 103534. https://doi.org/10.1016/j.inffus.2025.103534

    work page doi:10.1016/j.inffus.2025.103534 2025

  3. [3]

    B. Wang, F. Lin, M. Li, Z. Liang, Z. Chen, M. Wang, J.C. Cheng, Informative as-built modeling as a foundation for digital twins based on fine -grained object recognition and object -aware scan -vs-BIM for MEP scenes, Advanced Engineering Informatics 65 (2025) 103382. https://doi.org/10.1016/j.aei.2025.103382 29

    work page doi:10.1016/j.aei.2025.103382 2025

  4. [4]

    Schonberger, J

    J.L. Schonberger, J. -M. Frahm, Structure -from-motion revisited, Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 4104-4113. https://doi.org/10.1109/CVPR.2016.445

    work page doi:10.1109/cvpr.2016.445 2016

  5. [5]

    C. Yin, H. Yue, Q. Han, D. Hu, Z. Liang, F. Lin, B. Sun, B. Wang, M. Li, W. Yao, Industrial3D: A Terrestrial LiDAR Point Cloud Dataset and CrossParadigm Benchmark for Industrial Infrastructure, arXiv preprint arXiv:2603.28660 (2026). https://doi.org/10.48550/arXiv.2603.28660

    work page doi:10.48550/arxiv.2603.28660 2026

  6. [6]

    H. He, X. Li, G. Cheng, J. Shi, Y. Tong, G. Meng, V. Prinet, L. Weng, Enhanced boundary learning for glass - like object segmentation, Proceedings of the IEEE/CVF international conference on computer vision, 2021, pp. 15859-15868. https://doi.org/10.1109/ICCV48922.2021.01556

    work page doi:10.1109/iccv48922.2021.01556 2021

  7. [7]

    T. Yan, J. Gao, K. Xu, X. Zhu, H. Huang, H. Li, B. Wah, R.W. Lau, Ghostingnet: A novel approach for glass surface detection with ghosting cues, IEEE Transactions on Pattern Analysis and Machine Intelligence 47 (1) (2024) 323-337. https://doi.org/10.1109/TPAMI.2024.3463490

    work page doi:10.1109/tpami.2024.3463490 2024

  8. [8]

    J. Lin, Y. -H. Yeung, R. Lau, Exploiting semantic relations for glass surface detection, Advances in Neural Information Processing Systems 35 (2022) 22490-22504. https://openreview.net/forum?id=WrIrYMCZgbb

    2022

  9. [9]

    Chan, P.K

    J.C.F. Chan, P.K. -Y. WONG, X. Guo, Z. Liang, Y. Wu, J.C. Cheng, Enhancing Vision -Language Model for Construction Safety Inspection via Visually Grounded Reasoning and Reinforcement Learning, Available at SSRN 6294267. https://dx.doi.org/10.2139/ssrn.6294267

    work page doi:10.2139/ssrn.6294267

  10. [10]

    Mildenhall, P.P

    B. Mildenhall, P.P. Srinivasan, M. Tancik, J.T. Barron, R. Ramamoorthi, R. Ng, NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis, European Conference on Computer Vision, Springer, 2020, pp. 405-421. https://doi.org/10.1007/978-3-030-58452-8_24

    work page doi:10.1007/978-3-030-58452-8_24 2020

  11. [11]

    3D Gaussian Splatting for Real-Time Radiance Field Rendering,

    B. Kerbl, G. Kopanas, T. Leimkü hler, G. Drettakis, 3D Gaussian Splatting for Real -Time Radiance Field Rendering, ACM Transactions on Graphics 42 (4) (2023) 1-14. https://doi.org/10.1145/3592433

    work page doi:10.1145/3592433 2023

  12. [12]

    Z. Li, S. Yao, Y. Chu, A.F. Garcia -Fernandez, Y. Yue, W. Ding, X. Zhu, Mvg -splatting: Multi-view guided gaussian splatting with adaptive quantile-based geometric consistency densification, Information Fusion (2025) 103540. https://doi.org/10.1016/j.inffus.2025.103540

    work page doi:10.1016/j.inffus.2025.103540 2025

  13. [13]

    Wenliang Zhao, Lujia Bai, Yongming Rao, Jie Zhou, and Jiwen Lu

    Y. Jiang, J. Tu, Y. Liu, X. Gao, X. Long, W. Wang, Y. Ma, Gaussianshader: 3d gaussian splatting with shading functions for reflective surfaces, Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2024, pp. 5322-5332. https://doi.org/10.1109/CVPR52733.2024.00509

    work page doi:10.1109/cvpr52733.2024.00509 2024

  14. [14]

    Y. Yao, Z. Zeng, C. Gu, X. Zhu, L. Zhang, Reflective Gaussian Splatting, The Thirteenth International Conference on Learning Representations, 2025. https://openreview.net/forum?id=xPxHQHDH2u

    2025

  15. [15]

    K. Ye, Q. Hou, K. Zhou, 3d gaussian splatting with deferred reflection, ACM SIGGRAPH 2024 Conference Papers, 2024, pp. 1-10. https://doi.org/10.1145/3641519.3657456

    work page doi:10.1145/3641519.3657456 2024

  16. [16]

    C. Gu, X. Wei, Z. Zeng, Y. Yao, L. Zhang, Irgs: Inter-reflective gaussian splatting with 2d gaussian ray tracing, Proceedings of the Computer Vision and Pattern Recognition Conference, 2025, pp. 10943 -10952. https://doi.org/10.1109/CVPR52734.2025.01022

    work page doi:10.1109/cvpr52734.2025.01022 2025

  17. [17]

    Liang, Z

    Z. Liang, Z. Liang, M. Wang, L. Yang, J.C. Cheng, A Novel UAV View Planning Strategy for High Quality 3D Reconstruction of Irregular-Shaped Architecture, International Conference on Computing in Civil and Building Engineering, Springer, 2024, pp. 573-585. https://doi.org/10.1007/978-3-031-84208-5_42

    work page doi:10.1007/978-3-031-84208-5_42 2024

  18. [18]

    R. Jin, Y. Gao, Y. Wang, Y. Wu, H. Lu, C. Xu, F. Gao, Gs -planner: A gaussian -splatting-based planning framework for active high -fidelity reconstruction, 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), IEEE, 2024, pp. 11202-11209. https://doi.org/10.1109/IROS58592.2024.10801715

    work page doi:10.1109/iros58592.2024.10801715 2024

  19. [19]

    Matarneh, F

    S. Matarneh, F. Elghaish, F.P. Rahimian, N. Dawood, D. Edwards, Automated and interconnected facility management system: An open IFC cloud-based BIM solution, Automation in Construction 143 (2022) 104569. https://doi.org/10.1016/j.autcon.2022.104569

    work page doi:10.1016/j.autcon.2022.104569 2022

  20. [20]

    Huang, Z

    B. Huang, Z. Yu, A. Chen, A. Geiger, S. Gao, 2d gaussian splatting for geometrically accurate radiance fields, ACM SIGGRAPH 2024 conference papers, 2024, pp. 1-11. https://doi.org/10.1145/3641519.3657428

    work page doi:10.1145/3641519.3657428 2024

  21. [21]

    M. Ye, M. Danelljan, F. Yu, L. Ke, Gaussian grouping: Segment and edit anything in 3d scenes, European conference on computer vision, Springer, 2024, pp. 162-179. https://doi.org/10.1007/978-3-031-73397-0_10

    work page doi:10.1007/978-3-031-73397-0_10 2024

  22. [22]

    InProceedings of the SIGGRAPH Asia 2025 Conference Papers (SA Conference Papers ’25)

    A. Kirillov, E. Mintun, N. Ravi, H. Mao, C. Rolland, L. Gustafson, T. Xiao, S. Whitehead, A.C. Berg, W. -Y. Lo, Segment anything, Proceedings of the IEEE/CVF international conference on computer vision, 2023, pp. 4015-4026. https://doi.org/10.1109/ICCV51070.2023.00371

    work page doi:10.1109/iccv51070.2023.00371 2023

  23. [23]

    M. Qin, W. Li, J. Zhou, H. Wang, H. Pfister, Langsplat: 3d language gaussian splatting, Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 20051 -20060. https://doi.org/10.1109/CVPR52733.2024.01895 30

    work page doi:10.1109/cvpr52733.2024.01895 2024

  24. [24]

    Liang, J.C.F

    Z. Liang, J.C.F. Chan, J. Zhang, Z. Liang, B. Wang, M. Wang, J.C. Cheng, Multimodal LLM -driven language- embedded 3D gaussian splatting for semantic and realistic digitization of historical buildings, Automation in Construction 181 (2026) 106628. https://doi.org/10.1016/j.autcon.2025.106628

    work page doi:10.1016/j.autcon.2025.106628 2026

  25. [25]

    Radford, J.W

    A. Radford, J.W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clark, Learning transferable visual models from natural language supervision, International conference on machine learning, PmLR, 2021, pp. 8748-8763. https://proceedings.mlr.press/v139/radford21a.html

    2021

  26. [26]

    Caron, H

    M. Caron, H. Touvron, I. Misra, H. Jé gou, J. Mairal, P. Bojanowski, A. Joulin, Emerging properties in self - supervised vision transformers, Proceedings of the IEEE/CVF international conference on computer vision, 2021, pp. 9650-9660. https://doi.org/10.1109/ICCV48922.2021.00951

    work page doi:10.1109/iccv48922.2021.00951 2021

  27. [27]

    Zhang, J

    W. Zhang, J. Tang, W. Zhang, Y. Fang, Y. -S. Liu, Z. Han, MaterialRefGS: Reflective Gaussian Splatting with Multi-view Consistent Material Inference, The Thirty -ninth Annual Conference on Neural Information Processing Systems, 2025. https://openreview.net/forum?id=I64ZLbUP6u

    2025

  28. [28]

    J. Tang, F. Fei, Z. Li, X. Tang, S. Liu, Y. Chen, B. Huang, Z. Chen, X. Wu, B. Shi, SpecTRe -GS: Modeling Highly Specular Surfaces with Reflected Nearby Objects by Tracing Rays in 3D Gaussian Splatting, Proceedings of the Computer Vision and Pattern Recognition Conference, 2025, pp. 16133 -16142. https://doi.org/10.1109/CVPR52734.2025.01504

    work page doi:10.1109/cvpr52734.2025.01504 2025

  29. [29]

    Steering away from harm: An adaptive approach to defending vision language model against jailbreaks

    Y. Zhang, A. Chen, Y. Wan, Z. Song, J. Yu, Y. Luo, W. Yang, Ref-gs: Directional factorization for 2d gaussian splatting, Proceedings of the Computer Vision and Pattern Recognition Conference, 2025, pp. 26483 -26492. https://doi.org/10.1109/CVPR52734.2025.02466

    work page doi:10.1109/cvpr52734.2025.02466 2025

  30. [30]

    H. Mei, X. Yang, Y. Wang, Y. Liu, S. He, Q. Zhang, X. Wei, R.W. Lau, Don't hit me! glass detection in real - world scenes, Proceedings of the IEEE/cvf conference on computer vision and pattern recognition, 2020, pp. 3687-3696. https://doi.org/10.1109/CVPR42600.2020.00374

    work page doi:10.1109/cvpr42600.2020.00374 2020

  31. [31]

    K. Fan, C. Wang, Y. Wang, C. Wang, R. Yi, L. Ma, RFENet: towards reciprocal feature evolution for glass segmentation, Proceedings of the Thirty-Second International Joint Conference on Artificial Intelligence, 2023, pp. 717-725. https://doi.org/10.24963/ijcai.2023/80

    work page doi:10.24963/ijcai.2023/80 2023

  32. [32]

    J. Lin, Z. He, R.W. Lau, Rich context aggregation with reflection prior for glass surface detection, Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2021, pp. 13415 -13424. https://doi.org/10.1109/CVPR46437.2021.01321

    work page doi:10.1109/cvpr46437.2021.01321 2021

  33. [33]

    Gonzales, L.C

    M.E.M. Gonzales, L.C. Uy, J.P. Ilao, Designing a Lightweight Edge-Guided Convolutional Neural Network for Segmenting Mirrors and Reflective Surfaces, Proceedings of International Conference in Central Europe on Computer Graphics, Visualization and Computer Vision, 2023. https://www.doi.org/10.24132/CSRN.3301.14

    work page doi:10.24132/csrn.3301.14 2023

  34. [34]

    Hassen, M

    A.Y. Hassen, M. Arashpour, E. Abdi, Lightweight segmentation model for automated facade installation in high- rise buildings, Advanced Engineering Informatics 65 (2025) 103374. https://doi.org/10.1016/j.aei.2025.103374

    work page doi:10.1016/j.aei.2025.103374 2025

  35. [35]

    D. Wu, L. Cheng, R. Li, P. Yang, X. Xu, X. Wang, C. -H. Lee, Automatic segmentation of curtain wall frame using a context collaboration pyramid network, Engineering Applications of Artificial Intelligence 133 (2024) 108309. https://doi.org/10.1016/j.engappai.2024.108309

    work page doi:10.1016/j.engappai.2024.108309 2024

  36. [36]

    Y. Tian, F. Lin, Y. Li, T. Zhang, Q. Zhang, X. Fu, J. Huang, X. Dai, Y. Wang, C. Tian, UAVs meet LLMs: Overviews and perspectives towards agentic low -altitude mobility, Information Fusion 122 (2025) 103158. https://doi.org/10.1016/j.inffus.2025.103158

    work page doi:10.1016/j.inffus.2025.103158 2025

  37. [37]

    Liang, H

    Z. Liang, H. Li, H. Wu, P. Kok -Yiu Wong, F. Lin, Z. Liang, J.C. Cheng, A 3D Modeling Approach Based on UAV-Captured Low -Light Images, Computing in Civil Engineering 2024, pp. 680 -691. https://doi.org/10.1061/9780784486122.076

    work page doi:10.1061/9780784486122.076 2024

  38. [38]

    Liang, L

    Z. Liang, L. Yang, Z. Liang, J.C.F. Chan, Z. Zhang, M. Wang, J.C. Cheng, Optimized UAV view planning for high-quality 3D reconstruction of buildings using a modified sparrow search algorithm, Advanced Engineering Informatics 65 (2025) 103344. https://doi.org/10.1016/j.aei.2025.103344

    work page doi:10.1016/j.aei.2025.103344 2025

  39. [39]

    C. Peng, V. Isler, Adaptive view planning for aerial 3D reconstruction, 2019 International Conference on Robotics and Automation (ICRA), IEEE, 2019, pp. 2981-2987. https://doi.org/10.1109/ICRA.2019.8793532

    work page doi:10.1109/icra.2019.8793532 2019

  40. [40]

    Shang, Z

    Z. Shang, Z. Shen, Topology -based UAV path planning for multi -view stereo 3D reconstruction of complex structures, Complex & Intelligent Systems 9 (1) (2023) 909-926. https://doi.org/10.1007/s40747-022-00831-5

    work page doi:10.1007/s40747-022-00831-5 2023

  41. [41]

    Zheng, F

    X. Zheng, F. Wang, Z. Li, A multi-UAV cooperative route planning methodology for 3D fine-resolution building model reconstruction, ISPRS journal of photogrammetry and remote sensing 146 (2018) 483 -494. https://doi.org/10.1016/j.isprsjprs.2018.11.004

    work page doi:10.1016/j.isprsjprs.2018.11.004 2018

  42. [42]

    B. Hepp, M. Nieß ner, O. Hilliges, Plan3d: Viewpoint and trajectory optimization for aerial multi -view stereo reconstruction, ACM Transactions on Graphics (TOG) 38 (1) (2018) 1 -17. https://doi.org/10.1145/3233794 31

    work page doi:10.1145/3233794 2018

  43. [43]

    Zhang, C

    S. Zhang, C. Liu, N. Haala, Guided by model quality: UAV path planning for complete and precise 3D reconstruction of complex buildings, International Journal of Applied Earth Observation and Geoinformation 127 (2024) 103667. https://doi.org/10.1016/j.jag.2024.103667

    work page doi:10.1016/j.jag.2024.103667 2024

  44. [44]

    F. Wang, Y. Zou, E. del Rey Castillo, Y. Ding, Z. Xu, H. Zhao, J.B. Lim, Automated UAV path -planning for high-quality photogrammetric 3D bridge reconstruction, Structure and Infrastructure Engineering 20 (10) (2024) 1595-1614. https://doi.org/10.1080/15732479.2022.2152840

    work page doi:10.1080/15732479.2022.2152840 2024

  45. [45]

    X. Liu, W. Yi, P. Chen, Y. Tan, Flight path planning of UAV-driven refinement inspection for construction sites based on 3D reconstruction, Automation in Construction 177 (2025) 106360. https://doi.org/10.1016/j.autcon.2025.106360

    work page doi:10.1016/j.autcon.2025.106360 2025

  46. [46]

    Z. Xu, R. Jin, K. Wu, Y. Zhao, Z. Zhang, J. Zhao, F. Gao, Z. Gan, W. Ding, Hgs-planner: Hierarchical planning framework for active scene reconstruction using 3d gaussian splatting, 2025 IEEE International Conference on Robotics and Automation (ICRA), IEE E, 2025, pp. 14161 -14167. https://doi.org/10.1109/ICRA55743.2025.11127649

    work page doi:10.1109/icra55743.2025.11127649 2025

  47. [47]

    L. Jin, X. Zhong, Y. Pan, J. Behley, C. Stachniss, M. Popović, Activegs: Active scene reconstruction using gaussian splatting, IEEE Robotics and Automation Letters (2025). https://doi.org/10.1109/LRA.2025.3555149

    work page doi:10.1109/lra.2025.3555149 2025

  48. [48]

    Y. Li, Z. Kuang, T. Li, Q. Hao, Z. Yan, G. Zhou, S. Zhang, Activesplat: High -fidelity scene reconstruction through active gaussian splatting, IEEE Robotics and Automation Letters (2025). https://doi.org/10.1109/LRA.2025.3580331

    work page doi:10.1109/lra.2025.3580331 2025

  49. [49]

    C.-Y. Wu, Y. -C. Sun, C. -T. Lee, C. -H. Hsu, Optimally Planning Drone Trajectories to Capture 3D Gaussian Splatting Objects, International Conference on Multimedia Modeling, Springer, 2024, pp. 171 -185. https://doi.org/10.1007/978-981-96-2064-7_13

    work page doi:10.1007/978-981-96-2064-7_13 2024

  50. [50]

    Liang, X

    Z. Liang, X. Zhang, Z. Liang, B. Wang, M. Wang, J.C. Cheng, A Novel Framework for Generating Realistic and Semantic Digital Model of Reflective Glass Faç ades Using RefGlass -3DGS, Computing in Civil Engineering 2025: Computational and Intelligent Technolo gies, pp. 914 -923. https://doi.org/10.1061/9780784486436.098

    work page doi:10.1061/9780784486436.098 2025

  51. [51]

    Luká cs, R

    G. Luká cs, R. Martin, D. Marshall, Faithful least-squares fitting of spheres, cylinders, cones and tori for reliable segmentation, European conference on computer vision, Springer, 1998, pp. 671 -686. https://doi.org/10.1007/BFb0055697

    work page doi:10.1007/bfb0055697 1998

  52. [52]

    Galamhos, J

    C. Galamhos, J. Matas, J. Kittler, Progressive probabilistic Hough transform for line detection, Proceedings. 1999 IEEE computer society conference on computer vision and pattern recognition (Cat. No PR00149), Vol. 1, IEEE, 1999, pp. 554-560. https://doi.org/10.1109/CVPR.1999.786993

    work page doi:10.1109/cvpr.1999.786993 1999

  53. [53]

    Teeravech, M

    K. Teeravech, M. Nagai, K. Honda, M. Dailey, Discovering repetitive patterns in facade images using a RANSAC-style algorithm, ISPRS journal of photogrammetry and remote sensing 92 (2014) 38 -53. https://doi.org/10.1016/j.isprsjprs.2014.02.018

    work page doi:10.1016/j.isprsjprs.2014.02.018 2014

  54. [54]

    Wang, A.C

    Z. Wang, A.C. Bovik, H.R. Sheikh, E.P. Simoncelli, Image quality assessment: from error visibility to structural similarity, IEEE transactions on image processing 13 (4) (2004) 600 -612. https://doi.org/10.1109/TIP.2003.819861

    work page doi:10.1109/tip.2003.819861 2004

  55. [55]

    Maboudi, M

    M. Maboudi, M. Homaei, S. Song, S. Malihi, M. Saadatseresht, M. Gerke, A review on viewpoints and path planning for UAV -based 3-D reconstruction, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 16 (2023) 5026-5048. https://doi.org/10.1109/JSTARS.2023.3276427

    work page doi:10.1109/jstars.2023.3276427 2023

  56. [56]

    Kennedy, R

    J. Kennedy, R. Eberhart, Particle swarm optimization, Proceedings of ICNN'95 -international conference on neural networks, Vol. 4, ieee, 1995, pp. 1942-1948. https://doi.org/10.1109/ICNN.1995.488968

    work page doi:10.1109/icnn.1995.488968 1995

  57. [57]

    Rahaman, A

    N. Rahaman, A. Baratin, D. Arpit, F. Draxler, M. Lin, F. Hamprecht, Y. Bengio, A. Courville, On the spectral bias of neural networks, International conference on machine learning, PMLR, 2019, pp. 5301 -5310. https://proceedings.mlr.press/v97/rahaman19a.html

    2019

  58. [58]

    Tancik, P

    M. Tancik, P. Srinivasan, B. Mildenhall, S. Fridovich-Keil, N. Raghavan, U. Singhal, R. Ramamoorthi, J. Barron, R. Ng, Fourier features let networks learn high frequency functions in low dimensional domains, Advances in Neural Information Processing Syste ms 33 (2020) 7537 -7547. https://proceedings.neurips.cc/paper_files/paper/2020/file/55053683268957697...

    work page arXiv 2020

  59. [59]

    B. Wang, F. Lin, M. Li, Z. Liang, H. Yue, Q. Wang, J.C. Cheng, Aligning as-built and as-designed: Local point cloud to BIM registration via hybrid visibility map encoding for construction digital twins, Automation in Construction 180 (2025) 106551. https://doi.org/10.1016/j.autcon.2025.106551

    work page doi:10.1016/j.autcon.2025.106551 2025

  60. [60]

    Z. He, Y. -H. Wang, J. Zhang, Generative AIBIM: An automatic and intelligent structural design pipeline integrating BIM and generative AI, Information Fusion 114 (2025) 102654. https://doi.org/10.1016/j.inffus.2024.102654 32

    work page doi:10.1016/j.inffus.2024.102654 2025

  61. [61]

    B. Wang, Z. Chen, M. Li, Q. Wang, C. Yin, J.C. Cheng, Omni-Scan2BIM: A ready-to-use Scan2BIM approach based on vision foundation models for MEP scenes, Automation in Construction 162 (2024) 105384. https://doi.org/10.1016/j.autcon.2024.105384

    work page doi:10.1016/j.autcon.2024.105384 2024

  62. [62]

    M. Zhao, J. Chen, S. Song, Y. Li, F. Wang, S. Wang, D. Liu, Proposition of UAV multi-angle nap-of-the-object image acquisition framework based on a quality evaluation system for a 3D real scene model of a high -steep rock slope, International Journal of Ap plied Earth Observation and Geoinformation 125 (2023) 103558. https://doi.org/10.1016/j.jag.2023.103558

    work page doi:10.1016/j.jag.2023.103558 2023

  63. [63]

    Zhang, P

    R. Zhang, P. Isola, A.A. Efros, E. Shechtman, O. Wang, The unreasonable effectiveness of deep features as a perceptual metric, Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 586-595. https://doi.org/10.1109/CVPR.2018.00068

    work page doi:10.1109/cvpr.2018.00068 2018

  64. [64]

    TransmissiveGS: Residual-Guided Disentangled Gaussian Splatting for Transmissive Scene Reconstruction and Rendering

    Z. Liang, X. Zhang, T. Li, J.C. Cheng, C. -K. Tang, TransmissiveGS: Residual -Guided Disentangled Gaussian Splatting for Transmissive Scene Reconstruction and Rendering, arXiv preprint arXiv:2605.10705 (2026). https://doi.org/10.48550/arXiv.2605.10705

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2605.10705 2026

  65. [65]

    UAV-Assisted Scan-to-Simulation for Landslides Using Physics-Informed Gaussian Splatting

    Z. Liang, J.C. Cheng, UAV -Assisted Scan -to-Simulation for Landslides Using Physics -Informed Gaussian Splatting, arXiv preprint arXiv:2605.10715 (2026). https://doi.org/10.48550/arXiv.2605.10715

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2605.10715 2026