Reproduce “Mode-GS: Monocular Depth Guided Anchored 3D Gaussian Splatting for Robust Ground-View Scene Rendering” 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
Yonghan Lee, Jaehoon Choi, Dongki Jung, Jaeseong Yun, Soohyun Ryu, Dinesh Manocha, Suyong Yeon
Mode-GS builds renderable 3D scenes from robot images and poses without an input point cloud. It predicts monocular depth to place anchors, adjusts their scale, and generates Gaussian splats through residual decoding. On R3LIVE’s Campus scene, it reports PSNR 22.98 dB, SSIM 0.774, and LPIPS 0.365, ahead of LiDAR-input Scaffold-GS, but not on densely viewed Tanks and Temples.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.