Learning Task-Specific Generalized Convolutions in the Permutohedral Lattice

被引:1
|
作者
Wannenwetsch, Anne S. [1 ]
Kiefel, Martin [2 ]
Gehler, Peter V. [2 ]
Roth, Stefan [1 ]
机构
[1] Tech Univ Darmstadt, Darmstadt, Germany
[2] Amazon, Tubingen, Germany
来源
关键词
D O I
10.1007/978-3-030-33676-9_24
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Dense prediction tasks typically employ encoder-decoder architectures, but the prevalent convolutions in the decoder are not image-adaptive and can lead to boundary artifacts. Different generalized convolution operations have been introduced to counteract this. We go beyond these by leveraging guidance data to redefine their inherent notion of proximity. Our proposed network layer builds on the permutohedral lattice, which performs sparse convolutions in a high-dimensional space allowing for powerful non-local operations despite small filters. Multiple features with different characteristics span this permutohedral space. In contrast to prior work, we learn these features in a task-specific manner by generalizing the basic permutohedral operations to learnt feature representations. As the resulting objective is complex, a carefully designed framework and learning procedure are introduced, yielding rich feature embeddings in practice. We demonstrate the general applicability of our approach in different joint upsampling tasks. When adding our network layer to state-of-the-art networks for optical flow and semantic segmentation, boundary artifacts are removed and the accuracy is improved.
引用
收藏
页码:345 / 359
页数:15
相关论文
共 50 条
  • [1] Adversarial task-specific learning
    Fu, Xin
    Zhao, Yao
    Liu, Ting
    Wei, Yunchao
    Li, Jianan
    Wei, Shikui
    NEUROCOMPUTING, 2019, 362 : 118 - 128
  • [2] Task-specific disruption of perceptual learning
    Seitzt, AR
    Yamagishi, N
    Werner, B
    Goda, N
    Kawato, M
    Watanabe, T
    PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA, 2005, 102 (41) : 14895 - 14900
  • [3] Learning task-specific memory policies
    Rajendran, S
    Huber, M
    Proceedings of the Sixth IASTED International Conference on Intelligent Systems and Control, 2004, : 238 - 243
  • [4] A model of learning task-specific knowledge for a new task
    Taatgen, NA
    PROCEEDINGS OF THE TWENTY FIRST ANNUAL CONFERENCE OF THE COGNITIVE SCIENCE SOCIETY, 1999, : 730 - 735
  • [5] Task-Specific Automation in Deep Learning Processes
    Buchgeher, Georg
    Czech, Gerald
    Ribeiro, Adriano Souza
    Kloihofer, Werner
    Meloni, Paolo
    Busia, Paola
    Deriu, Gianfranco
    Pintor, Maura
    Biggio, Battista
    Chesta, Cristina
    Rinelli, Luca
    Solans, David
    Portela, Manuel
    DATABASE AND EXPERT SYSTEMS APPLICATIONS - DEXA 2021 WORKSHOPS, 2021, 1479 : 159 - 169
  • [6] Learning Task-Specific Strategies for Accelerated MRI
    Wu, Zihui
    Yin, Tianwei
    Sun, Yu
    Frost, Robert
    van der Kouwe, Andre
    Dalca, Adrian V.
    Bouman, Katherine L.
    IEEE TRANSACTIONS ON COMPUTATIONAL IMAGING, 2024, 10 : 1040 - 1054
  • [7] Learning Task-Specific City Region Partition
    Wang, Hongjian
    Jenkins, Porter
    Wei, Hua
    Wu, Fei
    Li, Zhenhui
    WEB CONFERENCE 2019: PROCEEDINGS OF THE WORLD WIDE WEB CONFERENCE (WWW 2019), 2019, : 3300 - 3306
  • [8] Learning to Compose Task-Specific Tree Structures
    Choi, Jihun
    Yoo, Kang Min
    Lee, Sang-goo
    THIRTY-SECOND AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE / THIRTIETH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE / EIGHTH AAAI SYMPOSIUM ON EDUCATIONAL ADVANCES IN ARTIFICIAL INTELLIGENCE, 2018, : 5094 - 5101
  • [9] Early motor learning and task-specific coordination
    Galloway, JC
    JOURNAL OF SPORT & EXERCISE PSYCHOLOGY, 2005, 27 : S11 - S11
  • [10] Probabilistic Learning of Task-Specific Visual Attention
    Borji, Ali
    Sihite, Dicky N.
    Itti, Laurent
    2012 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2012, : 470 - 477