# Anthropometric Landmarks Extraction and Dimensions Measurement Based on ResNet

^{*}

## Abstract

**:**

## 1. Introduction

- (1)
- A new way to extract landmarks from 2D images is proposed, that is, extracting landmarks by a deep convolutional neural network.
- (2)
- A multi-ellipse model is proposed, and the position of the axis division point of the ellipse model is determined by the thickness–width ratio of the body parts, which reduces the error of anthropometric dimensions.
- (3)
- The method is evaluated on real samples and compared with other methods, which shows the accuracy of the multi-ellipse model when the number of images is 2.

## 2. Related Work

## 3. Proposed Method

#### 3.1. Landmarks Extraction Based on ResNet

#### 3.2. Anthropometric Dimension Calculation Based on Multi-Ellipse Model

_{n}(0 < θ < 2π) of human body images and two sides of ellipse α

_{n}, β

_{n}. Assuming that the human body image has n angles, the elliptical segment at one of the angles is shown in Figure 5b.

_{1}, θ

_{2}is the rotation angle of human body image, C

_{1}, C

_{2}is the length of elliptical segment, p, q is the boundary points of elliptical segment C

_{2}, and α, β is the length of edge. A polar coordinate system ellipse equation with the center of ellipse as the origin is established, and the length of elliptical segment is calculated by the integral method as follows:

_{1}, αsinθ

_{2}), q(βcosθ

_{1}, βsinθ

_{2}). Due to the symmetry of the human body structure, only the length of curve segment in (0, π) is needed to be calculated.

_{all}represents dimension measurement and N represents rotation angle, which is the number of images.

_{1}= θ

_{2}= π/2, the key to this model lies in how to determine the position of the division point S. The traditional ellipse model uses a fixed division point and has a large error when facing people of different body shape. In this paper, the location of the division point is determined by thickness–width ratio of human body parts.

_{1}, b

_{1}] and the deformation range of the division point is [a

_{2}, b

_{2}], then at the thickness–width ratio c, the location of division point D should be:

## 4. Experiment and Result

#### 4.1. Datasets

#### 4.2. Training Details

#### 4.3. Result

_{q}represents the qth measurement, S

^{2}represents the sample variance. In this experiment, all researchers are required to make measurements twice.

_{i}) is evaluated as:

## 5. Discussion and Conclusions

## Author Contributions

## Funding

## Conflicts of Interest

## References

- Gültepe, U.; Güdükbay, U. Real-time virtual fitting with body measurement and motion smoothing. Comput. Graph.
**2014**, 43, 31–43. [Google Scholar] [CrossRef] - Mahnic, M.; Mahnica, M.; Petraka, S. Investigation of the Fit of Computer-based Parametric Garment Prototypes. J. Fiber Bioeng. Inform.
**2013**, 6, 51–61. [Google Scholar] [CrossRef] - Apeagyei, P.R.; Tyler, D. Ethical practice and methodological considerations in researching body cathexis for fashion products. J. Fash. Mark. Manag. Int. J.
**2007**, 11, 332–348. [Google Scholar] [CrossRef] - Heath, B.H.; Carter, J.E.L. A modified somatotype method. Am. J. Phys. Anthr.
**1967**, 27, 57–74. [Google Scholar] [CrossRef] [PubMed] - Ahn, S.H.; Park, J.; Nam, Y.-J.; Yun, M.H. 1C1-2 Analysis and Usability Testing of the 3D Scanning Method for Anthropometric Measurement of the Elderly. Jpn. J. Ergon.
**2015**, 51, S394–S397. [Google Scholar] [CrossRef] [Green Version] - Singh, A.K. The importance of anthropometric measurement error. Open Access Libr. J.
**2014**, 1, 1–9. [Google Scholar] [CrossRef] - Jain, S.; Rustagi, A.; Saurav, S.; Saini, R.; Singh, S. Three-dimensional CNN-inspired deep learning architecture for Yoga pose recognition in the real-world environment. Neural Comput. Appl.
**2020**, 1–15. [Google Scholar] [CrossRef] - Luvizon, D.C.; Picard, D.; Tabia, H. 2D/3D Pose Estimation and Action Recognition Using Multitask Deep Learning. In Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, UT, USA, 18–23 June 2018; pp. 5137–5146. [Google Scholar]
- Toshev, A.; Szegedy, C. DeepPose: Human Pose Estimation via Deep Neural Networks. In Proceedings of the 2014 IEEE Conference on Computer Vision and Pattern Recognition, Columbus, OH, USA, 23–28 June 2014; pp. 1653–1660. [Google Scholar]
- Arbelaez, P.; Maire, M.; Fowlkes, C.; Malik, J. Contour Detection and Hierarchical Image Segmentation. IEEE Trans. Pattern Anal. Mach. Intell.
**2011**, 33, 898–916. [Google Scholar] [CrossRef] [Green Version] - Chen, L.-C.; Papandreou, G.; Kokkinos, I.; Murphy, K.; Yuille, A.L. DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs. IEEE Trans. Pattern Anal. Mach. Intell.
**2017**, 40, 834–848. [Google Scholar] [CrossRef] [PubMed] - Dai, J.; He, K.; Sun, J. Instance-Aware Semantic Segmentation via Multi-task Network Cascades. In Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, USA, 27–30 June 2016; pp. 3150–3158. [Google Scholar]
- Aslam, M.; Rajbdad, F.; Khattak, S.; Azmat, S. Automatic measurement of anthropometric dimensions using frontal and lateral silhouettes. IET Comput. Vis.
**2017**, 11, 434–447. [Google Scholar] [CrossRef] - He, K.; Zhang, X.; Ren, S.; Sun, J. Deep Residual Learning for Image Recognition. In Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, USA, 27–30 June 2016. [Google Scholar]
- Cruz-Mota, J.; Bogdanova, I.; Paquier, B.; Bierlaire, M.; Thiran, J. Scale Invariant Feature Transform on the Sphere: Theory and Applications. Int. J. Comput. Vis.
**2012**, 98, 217–241. [Google Scholar] [CrossRef] - Bay, H.; Tuytelaars, T.; Van Gool, L. SURF: Speeded Up Robust Features. Comput. Vis. Image Underst.
**2008**, 110, 404–417. [Google Scholar] [CrossRef] - Loper, M.; Mahmood, N.; Romero, J.; Pons-Moll, G.; Black, M.J. SMPL: A Skinned Multi-Person Linear Model. ACM Trans. Graph. (TOG)
**2015**, 34, 1–16. [Google Scholar] [CrossRef] - Tompson, J.J.; Jain, A.; Lecun, Y.; Bregler, C. Joint Training of a Convolutional Network and a Graphical Model for Human Pose Estimation. arXiv
**2014**, arXiv:1406.2984. [Google Scholar] - Wei, S.-E.; Ramakrishna, V.; Kanade, T.; Sheikh, Y. Convolutional Pose Machines. In Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, USA, 27–30 June 2016; pp. 4724–4732. [Google Scholar]
- Nie, X.; Feng, J.; Zuo, Y.; Yan, S. Human Pose Estimation with Parsing Induced Learner. In Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, UT, USA, 18–23 June 2018; pp. 2100–2108. [Google Scholar]
- GB/T 38131-2019. Acquisition Method of Datum Points for Clothing Anthropometry; Standards Press of China: Beijing, China, 2019. [Google Scholar]
- GB/T 16160-2017. Anthropometric Definitions and Methods for Garment; Standards Press of China: Beijing, China, 2017. [Google Scholar]
- Konečný, J.; Richtárik, P. Semi-Stochastic Gradient Descent Methods. Front. Appl. Math. Stat.
**2017**, 3, 9. [Google Scholar] [CrossRef] [Green Version] - Pishchulin, L.; Insafutdinov, E.; Tang, S.; Andres, B.; Andriluka, M.; Gehler, P.; Schiele, B. DeepCut: Joint Subset Partition and Labeling for Multi Person Pose Estimation. In Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, USA, 27–30 June 2016; pp. 4929–4937. [Google Scholar]
- Lohman, T.G.; Roche, A.F.; Martorell, R. Anthropometric Standardization Reference Manual; Human Kinetics Books: Champaign, IL, USA, 1988. [Google Scholar]
- Lin, Y.-L.; Wang, M.-J.J. Constructing 3D human model from front and side images. Expert Syst. Appl.
**2012**, 39, 5012–5018. [Google Scholar] [CrossRef]

**Figure 1.**The overview of measurement algorithms. (

**a**) Input images; (

**b**) Convolutional Neural Network; (

**c**) Landmarks extraction; (

**d**) Anthropometric dimensions measurement.

**Figure 5.**The multi-ellipse model. (

**a**) Multi-ellipse model of waist; (

**b**) Ellipse segment in polar coordinate system.

**Figure 7.**The data of chest. (

**a**) deformation range of thickness–width ratio; (

**b**) deformation range of the division point. (

**c**) function of bust.

**Figure 9.**The agreement study of methods. (

**a**) study of A1 (Shoulder Breadth); (

**b**) study of C1 (Chest Girth).

**Figure 10.**The PA comparison of dimensions obtained by proposed method with other methods. (

**a**) comparison of dimensions (L1–A4); (

**b**) comparison of dimensions (B1–D3).

Landmark | Definition |
---|---|

FP1, SP1 Vertex | The highest point of the head. |

FP2 Right Acromion | The most lateral point of the lateral edge of the spine (acromial process) of the right scapula, projected vertically to the surface of the skin. |

FP3 Left Acromion | The most lateral point of the lateral edge of the spine (acromial process) of the left scapula, projected vertically to the surface of the skin. |

FP4 Right bust width point | The front fold point of the armpit on right arm scye line determined using a scale placed under arm. |

FP5 Left bust width point | The front fold point of the armpit on left arm scye line determined using a scale placed under arm. |

FP6 Anterior elbow point | The point on the anterior view of elbow. |

FP7 Right lateral waist point | The right intersection of the lateral midlines of human body and the waistline. |

FP8 Left lateral waist point | The left intersection of the lateral midlines of human body and the waistline. |

FP9 Right lateral buttock | The right intersection of the coronal plane of human body and the hip line. |

FP10 Left lateral buttock | The left intersection of the coronal plane of human body and the hip line. |

FP11 Right wrist point | The most prominent point of the bulge of the head of ulna in right arm. |

FP12 Perineum point or crotch | The intersection of the sagittal plane and the line connecting the lowest point of the left ischial tuberosity and right ischial tuberosity. |

SP2 Thelion | The most anterior point of the bust. |

SP3 Posterior bust depth point | The back fold point of the armpit on arm scye line determined using a scale placed under arm. |

SP4 Anterior arm depth point | The middle point on the anterior edge of arm. |

SP5 Posterior arm depth point | The middle point on the posterior edge of arm. |

SP6 Anterior waist point | The anterior intersection of the sagittal plane of human body and the waistline. |

SP7 Posterior waist point | The posterior intersection of the sagittal plane of human body and the waistline. |

SP8 Anterior peak of buttock | The most prominent point on the anterior edge of the ilium. |

SP9 Peak of buttock | The most prominent point in the buttock. |

SP10 Anterior thigh depth point | The middle point on the anterior edge of thigh. |

SP11 Posterior thigh depth point | The middle point on the posterior edge of thigh. |

SP12 Anterior crus depth point | The middle point on the anterior edge of crus. |

SP13 Posterior crus depth point | The middle point on the posterior edge of crus. |

SP14 Foot point | The inferior margin of prominence of the medial malleolus. |

Dimension | Definition | Dimension | Definition |
---|---|---|---|

L1 | Height | B2 | Waist Depth |

L2 | Upper arm Length | B3 | Hip Depth |

L3 | Lower arm Length | B4 | Thigh Depth |

L4 | Torso Height | B5 | Crus Depth |

L5 | Hip Height | C1 | Chest Girth |

L6 | Knee Height | C2 | Waist Girth |

A1 | Shoulder Breadth | C3 | Hip Girth |

A2 | Chest Breadth | D1 | Upper arm Girth |

A3 | Waist Breadth | D2 | Mid-Thigh Girth |

A4 | Hip Breadth | D3 | Calf Girth |

B1 | Bust Depth |

Number | Gender | Age | Weight (kg) | BMI |
---|---|---|---|---|

41 | female | 20–50 | 45–75 | 17–28 |

46 | male | 20–50 | 45–85 | 17–28 |

Training Times | Learning Rate |
---|---|

10,000 | 0.05 |

430,000 | 0.02 |

730,000 | 0.002 |

1,030,000 | 0.001 |

Image Type | Evaluation Method | Body Part | Landmark | Accuracy |
---|---|---|---|---|

Front image | PCK-0.5 | Shoulder * | FP2, FP3 | 97.80% |

Chest * | FP4, FP5 | 98.00% | ||

Waist * | FP7, FP8 | 97.40% | ||

Hip * | FP9, FP10 | 97.60% | ||

Head | FP1 | 98.20% | ||

Elbow | FP6 | 94.60% | ||

Wrist | FP11 | 93.20% | ||

Crotch | FP12 | 98.00% | ||

Knee | FP13 | 96.80% | ||

Ankle | FP14 | 97.20% | ||

Side image | PCK-0.5 | Arm * | SP4, SP5 | 93.20% |

Chest * | SP2, SP3 | 97.40% | ||

Waist * | SP6, SP7 | 97.40% | ||

Hip * | SP8, SP9 | 97.20% | ||

Thigh * | SP10, SP11 | 93.20% | ||

Calf * | SP12, SP13 | 93.60% | ||

Head | SP1 | 98.60% | ||

Ankle | SP14 | 97.20% |

Dimension | Intra-Observer Reliability | Inter-Observer Reliability | ||
---|---|---|---|---|

Trainer1 | Trainer2 | Trainer3 | R | |

L1 | 0.9 | 0.93 | 0.93 | 0.86 |

L2 | 0.87 | 0.89 | 0.88 | 0.83 |

L3 | 0.95 | 0.97 | 0.94 | 0.86 |

L4 | 0.85 | 0.93 | 0.91 | 0.82 |

L5 | 0.94 | 0.92 | 0.89 | 0.8 |

L6 | 0.97 | 0.98 | 0.95 | 0.92 |

A1 | 0.82 | 0.85 | 0.83 | 0.78 |

A2 | 0.88 | 0.92 | 0.93 | 0.82 |

A3 | 0.92 | 0.94 | 0.92 | 0.85 |

A4 | 0.83 | 0.85 | 0.81 | 0.84 |

B1 | 0.94 | 0.93 | 0.98 | 0.95 |

B2 | 0.87 | 0.9 | 0.91 | 0.91 |

B3 | 0.87 | 0.85 | 0.82 | 0.8 |

B4 | 0.86 | 0.89 | 0.88 | 0.85 |

B5 | 0.97 | 0.95 | 0.97 | 0.89 |

C1 | 0.9 | 0.93 | 0.91 | 0.84 |

C2 | 0.86 | 0.84 | 0.87 | 0.83 |

C3 | 0.97 | 0.96 | 0.96 | 0.92 |

D1 | 0.93 | 0.95 | 0.89 | 0.87 |

D2 | 0.86 | 0.89 | 0.88 | 0.85 |

D3 | 0.93 | 0.95 | 0.94 | 0.91 |

Type | Code | Body Part | MAD (mm) | MAE (mm) |
---|---|---|---|---|

height | L1 | height | 3.5 | 6 |

L2 | arm | 4.3 | 6 | |

L3 | forearm | 2.7 | 6 | |

L4 | back | 3.9 | 5 | |

L5 | pants | 5.8 | 7 | |

L6 | knee | 2 | 3 | |

width | A1 | shoulder | 5.1 | 8 |

A2 | chest | 3.4 | 8 | |

A3 | waist | 3.7 | 7 | |

A4 | hip | 4.2 | 7 | |

depth | B1 | chest | 2.2 | 4 |

B2 | waist | 2.5 | 4 | |

B3 | hip | 4.3 | 8 | |

B4 | thigh | 2.6 | 5 | |

B5 | calf | 2.2 | 5 | |

girth | D1 | arm | 7.3 | 9 |

D2 | thigh | 7.8 | 9 | |

D3 | calf | 7.5 | 9 | |

girth | C1 | chest | 6.4 | 15 |

C2 | waist | 6.9 | 11 | |

C3 | hip | 6.7 | 12 |

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Wang, X.; Liu, B.; Dong, Y.; Pang, S.; Tao, X.
Anthropometric Landmarks Extraction and Dimensions Measurement Based on ResNet. *Symmetry* **2020**, *12*, 1997.
https://doi.org/10.3390/sym12121997

**AMA Style**

Wang X, Liu B, Dong Y, Pang S, Tao X.
Anthropometric Landmarks Extraction and Dimensions Measurement Based on ResNet. *Symmetry*. 2020; 12(12):1997.
https://doi.org/10.3390/sym12121997

**Chicago/Turabian Style**

Wang, Xun, Baohua Liu, Yukun Dong, Shanchen Pang, and Xixi Tao.
2020. "Anthropometric Landmarks Extraction and Dimensions Measurement Based on ResNet" *Symmetry* 12, no. 12: 1997.
https://doi.org/10.3390/sym12121997