Obtaining accurate naked 3D body models and measurements in a clothed state is desirable for the apparel industry. In this paper, deep learning, body scanning, and millimeter-wave measurement techniques are combined to design a system that can accurately reconstruct a personalized naked human model even when heavy or loose clothing is worn. The system first acquires a frontal point cloud using depth cameras, then predicts the underclothing body shape through deep learning and registration with a parametric model. Millimeter-wave sensors are then utilized to precisely measure anthropometric parameters of key body parts, providing additional constraints for reconstruction. Finally, the parameters of the parametric model are optimized to match the measurements, obtaining a more accurate 3D reconstruction. Qualitative and quantitative experiments show that the mean and median error values of the reconstructed body key parts (chest circumference, waist circumference, hip circumference) are less than 2 cm, which gives the system an obvious advantage in reconstruction accuracy and robustness compared to deep learning methods alone. In addition, the reconstructed human body model is parametric, and the system can be applied not only in the apparel industry, but also has the potential in film and animation and other fields.
ChenZChenHOuyangY, et al.
A high-resolution and whole-body dataset of hand-object contact areas based on 3D scanning method. Sci Data2025;
12(1): 1–17.
2.
WangYXingCWeiYWuXZhengW. Single-view scene point cloud human grasp generation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 831–841.
3.
KangTHChoM.A comparative study of the sagittal and coronal balance of the spine measured using a 3D full-body scanner and whole spine X-ray. N Amer Spine Soc J2024;
18: 100386.
4.
BhatnagarBLSminchisescuCTheobaltCPons-MollG. Combining implicit function learning and parametric models for 3D human reconstruction. In Proceedings Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, 23–28 August 2020, Part II(16). Berlin: Springer, 2020, pp. 311–329.
5.
WangSGeigerATangS. Locally aware piecewise transformation fields for 3d human mesh registration. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021, pp. 7639–7648.
6.
JinkaSSSrivastavaAPokhariyaCSharmaANarayananPJ.Sharp: Shape-aware reconstruction of people in loose clothing. Int J Comput Vision2023;
131(4): 918–937.
7.
LiXLiGLiTLvJMitrouchevP.Design of a multi-sensor information acquisition system for mannequin reconstruction and human body size measurement under clothes. Text Res J2022;
92(19–20): 3750–3765.
8.
AnisuzzamanDMAl WalidMHSaifAS.Online trial room based on human body shape detection. Int J Image Graph Signal Process2019;
10(2): 21.
9.
TejedaYGMayerHA. A neural anthropometer learning from body dimensions computed on human 3D meshes. In 2021 IEEE Symposium Series on Computational Intelligence (SSCI). Piscataway, NJ: IEEE, 2021, pp. 1–8.
YanSWirtaJKämäräinenJ.Anthropometric clothing measurements from 3D body scans. Mach Vis Appl2020;
31(1–2): 7.
13.
XieHZhongY.Structure-consistent customized virtual mannequin reconstruction from 3D scans based on optimization. Text Res J2020;
90(7–8): 937–950.
14.
KaashkiNNHuPMunteanuA.Anet: A deep neural network for automatic 3D anthropometric measurement extraction. IEEE Trans Multimedia2021;
25: 831–844.
15.
ZhaoRDaiXHuPMunteanuA. W2H-Net: Fast prediction of waist-to-hip ratio from single partial dressed body scans in arbitrary postures via deep learning. In IEEE International Joint Conference on Biometrics. Piscataway, NJ: IEEE, 2023.
16.
LoperMMahmoodNRomeroJPons-MollGBlackMJ.SMPL: A skinned multi-person linear model. ACM Trans Graph2015;
34(6): 1–16.
17.
PavlakosGChoutasVGhorbaniN, et al. Expressive body capture: 3D hands, face, and body from a single image. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019, pp. 10975–10985.
18.
OsmanAABolkartTBlackMJ. Star: Sparse trained articulated human body regressor. In Proceedings of Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, 23–28 August 2020, Part VI(16). Berlin: Springer, 2020, pp. 598–613.
19.
WangYLiGZhangHZouXLiuYNieY.PanoMan: Sparse localized components–based model for full human motions. ACM Trans Graph2021;
40(2): 1–17.
20.
MullerLOsmanAATangSHuangCPBlackMJ. On self-contact and human pose. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021, pp. 9990–9999.
21.
ChoutasVMüllerLHuangCPTangSTzionasDBlackMJ. Accurate 3D body shape regression using metric and semantic attributes. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 2718–2728.
22.
ParkJJFlorencePStraubJNewcombeRLovegroveS. DeepSDF: Learning continuous signed distance functions for shape representation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019, pp. 165–174.
23.
MeschederLOechsleMNiemeyerMNowozinSGeigerA. Occupancy networks: Learning 3D reconstruction in function space. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019, pp. 4460–4470.
24.
ChibaneJAlldieckTPons-MollG. Implicit functions in feature space for 3d shape reconstruction and completion. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020, pp. 6970–6981.
25.
FengHKulitsPLiuSBlackMJAbrevayaVF. Generalizing neural human fitting to unseen poses with articulated SE(3) equivariance. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 7977–7988.
26.
LiXLiGLiMSongH.Parametric body reconstruction based on a single front scan point cloud. IEEE Trans Vis Comput Graph2024. doi: 10.1109/TVCG.2024.3475414.
27.
MaQYangJRanjanA, et al. Learning to dress 3D people in generative clothing. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020, pp. 6469–6478.
28.
ZhangCPujadesSBlackMJPons-MollG. Detailed, accurate, human shape estimation from clothed 3D scan sequences. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017, pp. 4191–4200.
29.
BhatnagarBLTiwariGTheobaltCPons-MollG. Multi-garment net: Learning to dress 3D people from images. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 2019, pp. 5420–5430.
30.
LoperMMBlackMJ. OpenDR: An approximate differentiable renderer. In Proceedings of Computer Vision–ECCV 2014: 13th European Conference, Zurich, Switzerland, 6–12 September 2014, Part VII(13). Berlin: Springer, 2014, pp. 154–169.
31.
ChibaneJPons-MollG. Implicit feature networks for texture completion from partial 3d data. In Proceedings of Computer Vision–ECCV 2020 Workshops, Glasgow, UK, 23–28 August 2020, Part II(16). Berlin:: Springer, 2020, pp. 717–725.
32.
LorensenWEClineHE. Marching cubes: A high resolution 3D surface construction algorithm. In Seminal Graphics: Pioneering Efforts that Shaped the Field. New York: ACM Press, 1998, pp. 347–353.
LiXLiGLiTMitrouchevP.Human body construction based on combination of parametric and nonparametric reconstruction methods. Visual Comput2023;
40(8): 5557–5573.
35.
CaoZSimonTWeiSSheikhY. Realtime multi-person 2D pose estimation using part affinity fields. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017, pp. 7291–7299.
36.
LiXLiGLiTLvJMitrouchevP.Remodeling of mannequins based on automatic binding of mesh to anthropometric parameters. Visual Comput2023;
39(12): 6435–6458.
37.
ISO. Size Designation of Clothes—Part 1: Anthropometric Definitions for Body Measurement. ISO Switzerland, 2017.
38.
ZhaoHJiangLJiaJTorrPHKoltunV. Point transformer. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 16259–16268.
39.
WuXLaoYJiangLLiuXZhaoH.Point transformer v2: Grouped vector attention and partition-based pooling. Adv Neur Inform Process Syst2022;
35: 33330–33342.
