{"ID":2835138,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2512.00385","arxiv_id":"2512.00385","title":"EZ-SP: Fast and Lightweight Superpoint-Based 3D Segmentation","abstract":"Superpoint-based pipelines provide an efficient alternative to point- or voxel-based 3D semantic segmentation, but are often bottlenecked by their CPU-bound partition step. We propose a learnable, fully GPU partitioning algorithm that generates geometrically and semantically coherent superpoints 13$\\times$ faster than prior methods. Our module is compact (under 60k parameters), trains in under 20 minutes with a differentiable surrogate loss, and requires no handcrafted features. Combine with a lightweight superpoint classifier, the full pipeline fits in $\u003c$2 MB of VRAM, scales to multi-million-point scenes, and supports real-time inference. With 72$\\times$ faster inference and 120$\\times$ fewer parameters, EZ-SP matches the accuracy of point-based SOTA models across three domains: indoor scans (S3DIS), autonomous driving (KITTI-360), and aerial LiDAR (DALES). Code and pretrained models are accessible at github.com/drprojects/superpoint_transformer.","short_abstract":"Superpoint-based pipelines provide an efficient alternative to point- or voxel-based 3D semantic segmentation, but are often bottlenecked by their CPU-bound partition step. We propose a learnable, fully GPU partitioning algorithm that generates geometrically and semantically coherent superpoints 13$\\times$ faster than...","url_abs":"https://arxiv.org/abs/2512.00385","url_pdf":"https://arxiv.org/pdf/2512.00385v2","authors":"[\"Louis Geist\",\"Loic Landrieu\",\"Damien Robert\"]","published":"2025-11-29T08:21:51Z","proceeding":"cs.CV","tasks":"[\"cs.CV\"]","methods":"[\"Transformer\"]","has_code":false}