Reproduce “Benchmarking Multi-View BEV Object Detection with Mixed Pinhole and Fisheye Cameras” 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
Xiangzhong Liu, Hao Shen
The authors evaluate 3D object detection using forward-facing stereo and two 180° fisheye views. On converted KITTI-360 data, they adapt view transformations with the MEI camera model and test polar coordinates. BEVDet’s mAP rises from 0.121 to 0.150, a reported 24% relative gain. PETR with polar coordinates reaches 0.280, below 0.295 for rectified images; rectification performs better beyond 10 m.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.