Reproduce “Sparse Recovery Bird's-Eye View Model Soup for Robust Camera-Based 3D Object Detection” 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
Jinwoo Chung, Sangho An, Hyunsung Park, Jangho Kim
SR-BEVS improves camera-based 3D detection by temporarily pruning low-magnitude weights in three Fast-BEV copies, fine-tuning each, and averaging recovered weights into one dense model. It is tested on nuScenes validation data with synthetic corruptions. Under 15% Gaussian noise, M0 NDS rises from 0.077 to 0.103, while clean-data mAP rises from 0.277 to 0.282; real-world performance is not established.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.