Learning Dynamic Hierarchical Models for Anytime Scene Labeling

Buyu Liu, Xuming He

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract / Description of output

With increasing demand for efficient image and video analysis, test-time cost of scene parsing becomes critical for many large-scale or time-sensitive vision applications. We propose a dynamic hierarchical model for anytime scene labeling that allows us to achieve flexible trade-offs between efficiency and accuracy in pixel-level prediction. In particular, our approach incorporates the cost of feature computation and model inference, and optimizes the model performance for any given test-time budget by learning a sequence of image-adaptive hierarchical models. We formulate this anytime representation learning as a Markov Decision Process with a discrete-continuous state-action space. A high-quality policy of feature and model selection is learned based on an approximate policy iteration method with action proposal mechanism. We demonstrate the advantages of our dynamic non-myopic anytime scene parsing on three semantic segmentation datasets, which achieves 90% of the state-of-the-art performances by using 15% of their overall costs.
Original languageEnglish
Title of host publicationComputer Vision - ECCV 2016 - 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, Part VI
PublisherSpringer
Pages650-666
Number of pages17
ISBN (Electronic)978-3-319-46466-4
ISBN (Print)978-3-319-46465-7
DOIs
Publication statusPublished - 17 Sept 2016
Externally publishedYes
Event14th European Conference on Computer Vision 2016 - Amsterdam, Netherlands
Duration: 8 Oct 201616 Oct 2016
http://www.eccv2016.org/

Publication series

NameLecture Notes in Computer Science
PublisherSpringer, Cham
Volume9910
ISSN (Print)0302-9743

Conference

Conference14th European Conference on Computer Vision 2016
Abbreviated titleECCV 2016
Country/TerritoryNetherlands
CityAmsterdam
Period8/10/1616/10/16
Internet address

Keywords / Materials (for Non-textual outputs)

  • Leaf Node
  • Hierarchical Model
  • Markov Decision Process
  • Feature Selection Method
  • Reward Function

Fingerprint

Dive into the research topics of 'Learning Dynamic Hierarchical Models for Anytime Scene Labeling'. Together they form a unique fingerprint.

Cite this