ASH: A Modern Framework for Parallel Spatial Hashing in 3D Perception

被引:5
作者
Dong, Wei [1 ]
Lao, Yixing [2 ]
Kaess, Michael [1 ,3 ]
Koltun, Vladlen [2 ,4 ]
机构
[1] Carnegie Mellon Univ, Robot Inst, Pittsburgh, PA 15213 USA
[2] Intel Corp, Intel Labs, Santa Clara, CA 95054 USA
[3] Georgia Tech, Coll Comp, Atlanta, GA 30332 USA
[4] Apple Inc, One Apple Pk Way, Cupertino, CA 95014 USA
关键词
Three-dimensional displays; Graphics processing units; Ash; Tensors; Task analysis; Image reconstruction; Arrays; Parallel hashing; GPU; volumetric reconstruction; SLAM; shape-from-shading; autodiff; RECONSTRUCTION;
D O I
10.1109/TPAMI.2022.3214347
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
We present ASH, a modern and high-performance framework for parallel spatial hashing on GPU. Compared to existing GPU hash map implementations, ASH achieves higher performance, supports richer functionality, and requires fewer lines of code (LoC) when used for implementing spatially varying operations from volumetric geometry reconstruction to differentiable appearance reconstruction. Unlike existing GPU hash maps, the ASH framework provides a versatile tensor interface, hiding low-level details from the users. In addition, by decoupling the internal hashing data structures and key-value data in buffers, we offer direct access to spatially varying data via indices, enabling seamless integration to modern libraries such as PyTorch. To achieve this, we 1) detach stored key-value data from the low-level hash map implementation; 2) bridge the pointer-first low level data structures to index-first high-level tensor interfaces via an index heap; 3) adapt both generic and non-generic integer-only hash map implementations as backends to operate on multi-dimensional keys. We first profile our hash map against state-of-the-art hash maps on synthetic data to show the performance gain from this architecture. We then show that ASH can consistently achieve higher performance on various large-scale 3D perception tasks with fewer LoC by showcasing several applications, including 1) point cloud voxelization, 2) retargetable volumetric scene reconstruction, 3) non-rigid point cloud registration and volumetric deformation, and 4) spatially varying geometry and appearance refinement. ASH and its example applications are open sourced in Open3D (http://www.open3d.org).
引用
收藏
页码:5417 / 5435
页数:19
相关论文
共 64 条
  • [1] Agarwal S., 2022, P ACM SIGGRAPH AS 20
  • [2] Alcantara DA, 2009, ACM T GRAPHIC, V28, DOI [10.1145/1618452.1618500, 10.1145/1618452.1610500]
  • [3] [Anonymous], 2007, INTEL TECHNOL J, V11, P309
  • [4] A Dynamic Hash Table for the GPU
    Ashkiani, Saman
    Farach-Colton, Martin
    Owens, John D.
    [J]. 2018 32ND IEEE INTERNATIONAL PARALLEL AND DISTRIBUTED PROCESSING SYMPOSIUM (IPDPS), 2018, : 419 - 429
  • [5] MULTIDIMENSIONAL BINARY SEARCH TREES USED FOR ASSOCIATIVE SEARCHING
    BENTLEY, JL
    [J]. COMMUNICATIONS OF THE ACM, 1975, 18 (09) : 509 - 517
  • [6] Bylow J., 2013, P ROB SCI SYST JUN 2, DOI [10.15607/RSS.2013.IX.035.[6]P, DOI 10.15607/RSS.2013.IX.035.[6]P]
  • [7] Celis P., 1985, 26th Annual Symposium on Foundations of Computer Science (Cat. No.85CH2224-4), P281, DOI 10.1109/SFCS.1985.48
  • [8] Chabra R., 2020, LNCS, V12374, P608, DOI 10.1007/978-3-030-58526-636
  • [9] Choi S, 2015, PROC CVPR IEEE, P5556, DOI 10.1109/CVPR.2015.7299195
  • [10] Deep Global Registration
    Choy, Christopher
    Dong, Wei
    Koltun, Vladlen
    [J]. 2020 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2020, : 2511 - 2520