Reproduce “RiO-DETRv2: Geometry-Aware Uncertainty Modeling Meets Real-Time Oriented Object Detection Transformer” 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
Zhangchi Hu, Zilong He, Yifan Zhao, Xiangyu Yu, Jie Chen, Peixi Wu, Yansong Peng, Hebei Li and 1 more
RiO-DETRv2 targets oriented object detection by refining side boundaries in an object-aligned frame, treating angle as a periodic quantity, and aligning classification with whole-box localization quality. On single-scale DOTA-1.0, RiO-DETRv2-x reaches 54.1 mAP50–95, compared with 52.6 for RiO-DETR-x, with nearly matched computation and measured patch-inference latency; this evidence does not establish full-pipeline timing.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.