Reproduce “AnyCrowd: Instance-Isolated Identity-Pose Binding for Arbitrary Multi-Character Animation” 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
Zhenyu Xie, Ji Xia, Michael Kampffmeyer, Panwen Hu, Zehua Ma, Yujian Zheng, Jing Wang, Zheng Chong and 4 more
AnyCrowd addresses pose-controlled group animation by separating people’s appearances before encoding and explicitly binding each identity to its pose track. Its attention coordinates identity-specific, background, and video information. On MCD-300’s self-driven test, it reaches 0.814 CLIP-I and 0.683 DINO-I, versus VACE’s 0.793 and 0.630. The density evidence covers held-out videos with 2–9 dancers, not arbitrary crowd sizes.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.