Self-Supervised Flow Estimation using Geometric Regularization with Applications to Camera Image and Grid Map Sequences

被引:0
|
作者
Wirges, Sascha [1 ]
Graeter, Johannes [2 ]
Zhang, Qiuhao [1 ]
Stiller, Christoph [2 ]
机构
[1] FZI Res Ctr Informat Technol, Mobile Percept Syst Grp, Karlsruhe, Germany
[2] KIT, Inst Measurement & Control Syst, Karlsruhe, Germany
关键词
D O I
暂无
中图分类号
U [交通运输];
学科分类号
08 ; 0823 ;
摘要
We present a self-supervised approach to estimate flow in camera image and top-view grid map sequences using fully convolutional neural networks in the domain of automated driving. We extend existing approaches for self-supervised optical flow estimation by adding a regularizer expressing motion consistency assuming a static environment. However, as this assumption is violated for other moving traffic participants we also estimate a mask to scale this regularization. Adding a regularization towards motion consistency improves convergence and flow estimation accuracy. Furthermore, we scale the errors due to spatial flow inconsistency by a mask that we derive from the motion mask. This improves accuracy in regions where the flow drastically changes due to a better separation between static and dynamic environment. We apply our approach to optical flow estimation from camera image sequences, validate on odometry estimation and suggest a method to iteratively increase optical flow estimation accuracy using the generated motion masks. Finally, we provide quantitative and qualitative results based on the KITTI odometry and tracking benchmark for scene flow estimation based on grid map sequences. We show that we can improve accuracy and convergence when applying motion and spatial consistency regularization.
引用
收藏
页码:1782 / 1787
页数:6
相关论文
共 50 条
  • [41] Few-shot adaptation of GANs using self-supervised consistency regularization
    Israr, Syed Muhammad
    Saeed, Rehan
    Zhao, Feng
    KNOWLEDGE-BASED SYSTEMS, 2024, 302
  • [42] Learning by Distillation: A Self-Supervised Learning Framework for Optical Flow Estimation
    Liu, Pengpeng
    Lyu, Michael R.
    King, Irwin
    Xu, Jia
    IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2022, 44 (09) : 5026 - 5041
  • [43] SurroundDepth: Entangling Surrounding Views for Self-Supervised Multi-Camera Depth Estimation
    Wei, Yi
    Zhao, Linqing
    Zheng, Wenzhao
    Zhu, Zheng
    Rao, Yongming
    Huang, Guan
    Lu, Jiwen
    Zhou, Jie
    CONFERENCE ON ROBOT LEARNING, VOL 205, 2022, 205 : 539 - 549
  • [44] Rigid-aware self-supervised GAN for camera ego-motion estimation
    Lin, Lili
    Luo, Wan
    Yan, Zhengmao
    Zhou, Wenhui
    DIGITAL SIGNAL PROCESSING, 2022, 126
  • [45] Self-Supervised SAR Despeckling Using Deep Image Prior
    Albisani, Chiara
    Baracchi, Daniele
    Piva, Alessandro
    Argenti, Fabrizio
    PATTERN RECOGNITION LETTERS, 2025, 190 : 169 - 176
  • [46] UnRectDepthNet: Self-Supervised Monocular Depth Estimation using a Generic Framework for Handling Common Camera Distortion Models
    Kumar, Varun Ravi
    Yogamani, Senthil
    Bach, Markus
    Witt, Christian
    Milz, Stefan
    Maeder, Patrick
    2020 IEEE/RSJ INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS (IROS), 2020, : 8177 - 8183
  • [47] FisheyeDistanceNet: Self-Supervised Scale-Aware Distance Estimation using Monocular Fisheye Camera for Autonomous Driving
    Kumar, Varun Ravi
    Hiremath, Sandesh Athni
    Bach, Markus
    Milz, Stefan
    Witt, Christian
    Pinard, Clement
    Yogamani, Senthil
    Maeder, Patrick
    2020 IEEE INTERNATIONAL CONFERENCE ON ROBOTICS AND AUTOMATION (ICRA), 2020, : 574 - 581
  • [48] Depth360: Self-supervised Learning for Monocular Depth Estimation using Learnable Camera Distortion Model
    Hirose, Noriaki
    Tahara, Kosuke
    2022 IEEE/RSJ INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS (IROS), 2022, : 317 - 324
  • [49] Self-Supervised Monocular Depth Estimation With Geometric Prior and Pixel-Level Sensitivity
    Liu, Jierui
    Cao, Zhiqiang
    Liu, Xilong
    Wang, Shuo
    Yu, Junzhi
    IEEE TRANSACTIONS ON INTELLIGENT VEHICLES, 2023, 8 (03): : 2244 - 2256
  • [50] STViT: Improving Self-Supervised Multi-Camera Depth Estimation with Spatial-Temporal Context and Adversarial Geometry Regularization (Student Abstract)
    Chen, Zhuo
    Zhao, Haimei
    Yuan, Bo
    Li, Xiu
    THIRTY-EIGTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, VOL 38 NO 21, 2024, : 23460 - 23461