Reproduce “Human Motion Prediction via Continual Prior Compensation” 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
Jianwei Tang, Jian-Fang Hu, Tianming Liang, Xiaotong Lin, Jiangxin Sun, Wei-Shi Zheng, Jianhuang Lai
CPC trains human-motion predictors progressively on successive future segments and learns a Prior Compensation Factor (PCF) to adjust for loss of earlier information; CPC++ adds a fine-grained factor for separate prediction subtasks. On benchmark datasets and tested backbones, the methods generally improve or tie PGBIG, including 110.3 to 108.8 MPJPE at 1000 ms on Human3.6M, not evidence of universal transfer.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.