Autoregressive Uncertainty Modeling for 3D Bounding Box Prediction

被引:1
|
作者
Liu, YuXuan [1 ,2 ]
Mishra, Nikhil [1 ,2 ]
Sieb, Maximilian [1 ]
Shentu, Yide [1 ,2 ]
Abbeel, Pieter [1 ,2 ]
Chen, Xi [1 ]
机构
[1] Covariant, Emeryville, CA 94608 USA
[2] Univ Calif Berkeley, Berkeley, CA 94720 USA
来源
关键词
3D bounding boxes; 3D bounding box estimation; 3D object detection; Autoregressive models; Uncertainty modeling;
D O I
10.1007/978-3-031-20080-9_39
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
3D bounding boxes are a widespread intermediate representation in many computer vision applications. However, predicting them is a challenging task, largely due to partial observability, which motivates the need for a strong sense of uncertainty. While many recent methods have explored better architectures for consuming sparse and unstructured point cloud data, we hypothesize that there is room for improvement in the modeling of the output distribution and explore how this can be achieved using an autoregressive prediction head. Additionally, we release a simulated dataset, COB-3D, which highlights new types of ambiguity that arise in real-world robotics applications, where 3D bounding box prediction has largely been underexplored. We propose methods for leveraging our autoregressive model to make high confidence predictions and meaningful uncertainty measures, achieving strong results on SUN-RGBD, Scannet, KITTI, and our new dataset.
引用
收藏
页码:673 / 694
页数:22
相关论文
共 50 条
  • [31] Uncertainty Estimation of the Bounding Box Centroid for Autonomous Driving
    Yang, Kaipei
    Ye, Shida
    Bar-Shalom, Yaakov
    Hunt, Shawn
    [J]. AUTONOMOUS SYSTEMS: SENSORS, PROCESSING AND SECURITY FOR GROUND, AIR, SEA AND SPACE VEHICLES AND INFRASTRUCTURE 2022, 2022, 12115
  • [32] Tora3D: an autoregressive torsion angle prediction model for molecular 3D conformation generation
    Zhang, Zimei
    Wang, Gang
    Li, Rui
    Ni, Lin
    Zhang, RunZe
    Cheng, Kaiyang
    Ren, Qun
    Kong, Xiangtai
    Ni, Shengkun
    Tong, Xiaochu
    Luo, Li
    Wang, Dingyan
    Lu, Xiaojie
    Zheng, Mingyue
    Li, Xutong
    [J]. JOURNAL OF CHEMINFORMATICS, 2023, 15 (01)
  • [33] Behavior Prediction Using 3D Box Estimation in Road Environment
    Kaida, Shinnosuke
    Kiawjak, Pornprom
    Matsushima, Kousuke
    [J]. 2020 5TH INTERNATIONAL CONFERENCE ON COMPUTER AND COMMUNICATION SYSTEMS (ICCCS 2020), 2020, : 256 - 260
  • [34] Tora3D: an autoregressive torsion angle prediction model for molecular 3D conformation generation
    Zimei Zhang
    Gang Wang
    Rui Li
    Lin Ni
    RunZe Zhang
    Kaiyang Cheng
    Qun Ren
    Xiangtai Kong
    Shengkun Ni
    Xiaochu Tong
    Li Luo
    Dingyan Wang
    Xiaojie Lu
    Mingyue Zheng
    Xutong Li
    [J]. Journal of Cheminformatics, 15
  • [35] Novel Vehicle Bounding Box Tracking Using a Low-End 3D Laser Scanner
    An, Jhonghyun
    Kim, Euntai
    [J]. IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS, 2021, 22 (06) : 3403 - 3419
  • [36] Pre-Impact Fall Detection System Using Dynamic Threshold and 3D Bounding Box
    Otanasap, Nuth
    Boonbrahm, Poonpong
    [J]. EIGHTH INTERNATIONAL CONFERENCE ON GRAPHIC AND IMAGE PROCESSING (ICGIP 2016), 2017, 10225
  • [37] Application of uncertainty modeling in 2D and 3D object detection
    Wang, Meng
    Zhu, Bing
    [J]. Xi Tong Gong Cheng Yu Dian Zi Ji Shu/Systems Engineering and Electronics, 2023, 45 (08): : 2370 - 2376
  • [38] 3D gravity in a box
    Kraus, Per
    Monten, Ruben
    Myers, Richard M.
    [J]. SCIPOST PHYSICS, 2021, 11 (03):
  • [39] Monocular 3D Object Detection Based on Uncertainty Prediction of Keypoints
    Chen, Mu
    Zhao, Huaici
    Liu, Pengfei
    [J]. MACHINES, 2022, 10 (01)
  • [40] 3D data driven prediction for active contour models based on geometric bounding volumes
    Denzler, J
    Niemann, H
    [J]. PATTERN RECOGNITION LETTERS, 1996, 17 (11) : 1171 - 1178