Learning Anthropometry from Rendered Humans

Abstract

Accurate estimation of anthropometric body measurements from RGB imags has many potential applications in industrial design, online clothing, medical diagnosis and ergonomics. Research on this topic is limited by the fact that there exist only generated datasets which are based on fitting a 3D body mesh to 3D body scans in the commercial CAESAR dataset. For 2D only silhouettes are generated. To circumvent the data bottleneck, we introduce a new 3D scan dataset of 2,675 female and 1,474 male scans. We also introduce a small dataset of 200 RGB images and tape measured ground truth. With the help of the two new datasets we propose a part-based shape model and a deep neural network for estimating anthropometric measurements from 2D images. All data will be made publicly available.

Linear Regression vs. Deep Learning: A Simple Yet Effective Baseline for Human Body Measurement

“…Image-based approaches (2D) can be divided into shape-aware pose estimation methods, which typically regress pose and shape parameters in-the-wild either from 2D keypoints or directly from images [10,11,[13][14][15][16][17][52][53][54][55][56][57][58][59][60][61][62][63][64][65][66], and shape estimation methods, which regress shape from silhouettes, usually in fixed pose and minimal clothing [21][22][23][24][25][26][27][28]31,67]. We compare the proposed baseline against the state-of-the-art 3D-and 2D-based approaches for human body measurement estimation and achieve comparable performance to the best methods, while outperforming several deep learning models (see Section 4).…”

“…We use a total of 18 body measurements, 15 of which are a standard set of measurements used in previous works [21][22][23][24][25][26][27][28]31,67], and 3 of which are used specifically to compare with Virtual Caliper [31] (see Table 1). The measurements are either lengths or circumferences and are calcuated using their corresponding landmarks.…”