Core Concepts
Language-guided scene-aware human motion generation dataset, LaserHuman, revolutionizes research with diverse motions and free-form language descriptions.
Abstract
LaserHuman dataset introduced for scene-text-to-motion research.
Includes real human motions in 3D scenes with diverse scenarios.
Multi-conditional diffusion model proposed for semantically consistent and physically plausible motion generation.
Comparison with existing datasets and methods for evaluation.
Stats
Figure 1. LaserHuman consists of large-scale sequences of rich human motions and abundant human interactions captured in various real scenarios with free-form language descriptions.
LaserHuman contains 11 diverse 3D scenes, 3,374 high-quality motion sequences, and 12,303 language descriptions.