Single-shot 3D multi-person pose estimation in complex images
This addresses the problem of estimating 3D poses for multiple people in cluttered scenes for computer vision applications, representing a strong incremental advance.
The paper tackles 3D multi-person pose estimation in complex images with a single-shot method that jointly locates joints, estimates 3D coordinates, and groups them into skeletons without needing bounding boxes. It significantly outperforms state-of-the-art methods on CMU Panoptic and MuPoTS-3D datasets, and shows good results on the JTA Dataset.
In this paper, we propose a new single shot method for multi-person 3D human pose estimation in complex images. The model jointly learns to locate the human joints in the image, to estimate their 3D coordinates and to group these predictions into full human skeletons. The proposed method deals with a variable number of people and does not need bounding boxes to estimate the 3D poses. It leverages and extends the Stacked Hourglass Network and its multi-scale feature learning to manage multi-person situations. Thus, we exploit a robust 3D human pose formulation to fully describe several 3D human poses even in case of strong occlusions or crops. Then, joint grouping and human pose estimation for an arbitrary number of people are performed using the associative embedding method. Our approach significantly outperforms the state of the art on the challenging CMU Panoptic and a previous single shot method on the MuPoTS-3D dataset. Furthermore, it leads to good results on the complex and synthetic images from the newly proposed JTA Dataset.