“XNect: real-time multi-person 3D motion capture with a single RGB camera” by Mehta, Sotnychenko, Mueller, Xu, Elgharib, et al. …
Notice: Pod Template PHP code has been deprecated, please use WP Templates instead of embedding PHP. has been deprecated since Pods version 2.3 with no alternative available. in /data/siggraph/websites/history/wp-content/plugins/pods/includes/general.php on line 518
Conference:
- SIGGRAPH 2020
-
More from SIGGRAPH 2020:
Notice: Array to string conversion in /data/siggraph/websites/history/wp-content/plugins/siggraph-archive-plugin/src/next_previous/source.php on line 345
Notice: Array to string conversion in /data/siggraph/websites/history/wp-content/plugins/siggraph-archive-plugin/src/next_previous/source.php on line 345
Type(s):
Title:
- XNect: real-time multi-person 3D motion capture with a single RGB camera
Presenter(s):
Description:
We present a real-time approach for multi-person 3D motion capture at over 30 fps using a single RGB camera. It operates successfully in generic scenes which may contain occlusions by objects and by other people. Our method operates in subsequent stages. The first stage is a convolutional neural network (CNN) that estimates 2D and 3D pose features along with identity assignments for all visible joints of all individuals. We contribute a new architecture for this CNN, called SelecSLS Net, that uses novel selective long and short range skip connections to improve the information flow allowing for a drastically faster network without compromising accuracy. In the second stage, a fullyconnected neural network turns the possibly partial (on account of occlusion) 2D pose and 3D pose features for each subject into a complete 3D pose estimate per individual. The third stage applies space-time skeletal model fitting to the predicted 2D and 3D pose per subject to further reconcile the 2D and 3D pose, and enforce temporal coherence. Our method returns the full skeletal pose in joint angles for each subject. This is a further key distinction from previous work that do not produce joint angle results of a coherent skeleton in real time for multi-person scenes. The proposed system runs on consumer hardware at a previously unseen speed of more than 30 fps given 512×320 images as input while achieving state-of-the-art accuracy, which we will demonstrate on a range of challenging real-world scenes.
