Reproduce “CRAFT: Camera-Radar 3D Object Detection with Spatio-Contextual Fusion 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

Youngseok Kim, Sanmin Kim, Jun Won Choi, Dongsuk Kum

CRAFT uses camera–radar fusion to refine camera 3D object proposals. Soft Polar Association gathers nearby radar points using proposal uncertainty, and an attention-based transformer exchanges visual and spatial features. A fusion confidence score determines whether to update the camera estimate or retain it. On the nuScenes test set, CRAFT reports 41.1 mAP and 52.3 NDS, versus CenterFusion’s 32.6 mAP and 44.9 NDS.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.