ARCH: Animatable Reconstruction of Clothed Humans

We propose a novel, end-to-end pipeline to reconstruct detailed animatable avatars from a single picture.

Chained Representation Cycling: Learning to Estimate 3D Human Pose and Shape by Cycling Between Representations

Extending cycle consistency models to learn mappings between 2D images and parameterized 3D models.

Efficient Learning on Point Clouds With Basis Point Sets

We propose a novel representation, Basis Point Sets (BPS), for point clouds that allows us to perform deep learning tasks naturally and efficiently.

Neural Body Fitting: Unifying Deep Learning and Model-Based Human Pose and Shape Estimation

We propose a novel end-to-end neural network architecture and training scheme to train 3D body pose and shape estimation from 2D images.

Towards Accurate Marker-less Human Shape and Pose Estimation over Time

Existing markerless motion capture methods often assume known backgrounds, static cameras, and sequence specific motion priors, limiting their application scenarios. Here we present a fully automatic method that, given multiview videos, estimates 3D …

A Generative Model of People in Clothing

We propose an end-to-end trainable neural network that can generate images of people in clothing. It can be conditioned on pose, body shape and clothing color.

Unite the People: Closing the Loop Between 3D and 2D Human Representations

We propose an annotation pipeline and models for detailed 3D human body model fits to 2D images. In this paper, we explore models with up to 91 keypoints and 32 semantic body part segments.

Early Stopping Without a Validation Set

We propose a strategy to assess overfitting using gradient and weight statistics in neural networks. This makes the use of a separate validation set unnecessary.

Barrista - Caffe Well-Served

The caffe framework is one of the leading deep learning toolboxes in the machine learning and computer vision community. While it offers efficiency and configurability, it falls short of a full interface to Python. With increasingly involved …

Keep it SMPL: Automatic Estimation of 3D Human Pose and Shape from a Single Image

We describe the first method to automatically estimate the 3D pose of the human body as well as its 3D shape from a single unconstrained image. We estimate a full 3D mesh and show that 2D joints alone carry a surprising amount of information about …