Reproduce “Event-T2M: Event-level Conditioning for Complex Text-to-Motion Synthesis” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Seong-Eun Hong, JaeYoung Seon, JuYeong Hwang, JongHwan Shin, HyeongYeop Kang
Event-T2M generates motion from multi-event text prompts by segmenting prompts into meaningful action units, encoding each with TMR, and conditioning generation on event-level and global representations. In a matched TMR comparison on HumanML3D-E prompts with at least four events, event-level conditioning raises Top-1 R-Precision from 0.440 to 0.466 and lowers FID from 0.355 to 0.265; this result does not guarantee correct ordering for every prompt.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.