Reproduce “DeGuNet: Depth-Guided Ultra-Compact Backbones for Efficient LiDAR-Camera 3D 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
Haifa Zhang, Yijing Wang, Peixi Peng, Zhiqiang Zuo
DeGuNet replaces the image backbone in LiDAR–camera 3D detection with a compact encoder pretrained for depth completion from images and sparse LiDAR. Its sparsity-aware operations use valid LiDAR regions, and its pretraining decoder is discarded. On KITTI, it reports 69.40 detection mAP versus ResNet’s 62.11 at similar size; its minimalist design may limit rare-category semantics.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.