3D Human Motion
3D human motion research focuses on accurately capturing, generating, and manipulating human movement in three dimensions. Current efforts concentrate on developing robust methods for recovering 3D motion from various input modalities (e.g., video, IMU data), employing advanced architectures like diffusion models, transformers, and neural operators to generate realistic and diverse motions, often conditioned on text or other contextual information. This field is crucial for advancements in areas such as virtual reality, animation, robotics, and human-computer interaction, enabling more natural and intuitive interactions with digital environments and machines.
Papers
Markerless 3D human pose tracking through multiple cameras and AI: Enabling high accuracy, robustness, and real-time performance
Luca Fortini, Mattia Leonori, Juan M. Gandarias, Elena de Momi, Arash Ajoudani
CIRCLE: Capture In Rich Contextual Environments
Joao Pedro Araujo, Jiaman Li, Karthik Vetrivel, Rishi Agarwal, Deepak Gopinath, Jiajun Wu, Alexander Clegg, C. Karen Liu
PaCMO: Partner Dependent Human Motion Generation in Dyadic Human Activity using Neural Operators
Md Ashiqur Rahman, Jasorsi Ghosh, Hrishikesh Viswanath, Kamyar Azizzadenesheli, Aniket Bera
FutureHuman3D: Forecasting Complex Long-Term 3D Human Behavior from Video Observations
Christian Diller, Thomas Funkhouser, Angela Dai