Full-body 3D scanning simplifies the acquisition of digital body models. However, current systems are bulky, intricate, and costly, with strict clothing constraints. We propose a pipeline that combines inner body shape inference and parametric model registration for reconstructing the corresponding body model from a single front scan of a clothed body. Three networks modules (Scan2Front-Net, Front2Back-Net, and Inner2Corr-Net) with relatively independent functions are proposed for predicting front inner, back inner, and parametric model reference point clouds, respectively. We consider the back inner point cloud as an axial offset of the front inner point cloud and divide the body into 14 parts. This offset relationship is then learned within the same body parts to reduce the ambiguity of the inference. The predicted front and back inner point clouds are concatenated as inner body point cloud, and then reconstruction is achieved by registering the parametric body model through a point-to-point correspondence between the reference point cloud and the inner body point cloud. Qualitative and quantitative analysis show that the proposed method has significant advantages in terms of body shape completion and reconstruction body model accuracy.

In this paper we propose a system to provide automatic location of objects (object segmentation) in video sequences using motion and color statistics at quasi real time performance. The proposed algorithm uses both color and motion information to generate two independent models of the scene: color model and motion model. Both models are combined into a cost function that encodes the likelihood of a pixel to either belonging to the object or not. The algorithm works in a two-stage process: (1) Motion model segmentation and (2) Color model contribution. In case the intermediate results obtained at the end of the first stage (motion model segmentation) are consistent in comparison to previous frame results, stage two (color model) is skipped, allowing us to achieve 15 fps at HD resolutions, while maintaining segmentation quality. We also show an application of video object segmentation by applying our algorithm to create panoramic video summaries.