Reproduce “You Only Learn One Query: Learning Unified Human Query for Single-Stage Multi-Person Multi-Task Human-Centric Perception” 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
Sheng Jin, Shuhuai Li, Tong Li, Wentao Liu, Chen Qian, Ping Luo
HQNet provides a single-stage system for detecting people and predicting multiple human-centric outputs together. It uses a shared Human Query representation for each person, with task-specific heads for boxes, poses, masks, attributes, and meshes. On OCHuman’s test set, HQNet with ResNet-50 reaches 31.1 segmentation AP, compared with 20.1 for CondInst; this comparison does not establish that the query alone caused the difference.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.