English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

This paper formulates and presents a solution to the new problem of budgetedsemantic video segmentation. Given a video, the goal is to accurately assign asemantic class label to every pixel in the video within a specified timebudget. Typical approaches to such labeling problems, such as ConditionalRandom Fields (CRFs), focus on maximizing accuracy but do not provide aprincipled method for satisfying a time budget. For video data, the timerequired by CRF and related methods is often dominated by the time to computelow-level descriptors of supervoxels across the video. Our key contribution isthe new budgeted inference framework for CRF models that intelligently selectsthe most useful subsets of descriptors to run on subsets of supervoxels withinthe time budget. The objective is to maintain an accuracy as close as possibleto the CRF model with no time bound, while remaining within the time budget.Our second contribution is the algorithm for learning a policy for the sparseselection of supervoxels and their descriptors for budgeted CRF inference. Thislearning algorithm is derived by casting our problem in the framework of MarkovDecision Processes, and then instantiating a state-of-the-art policy learningalgorithm known as Classification-Based Approximate Policy Iteration. Ourexperiments on multiple video datasets show that our learning approach andframework is able to significantly reduce computation time, and maintaincompetitive accuracy under varying budgets.

Approximate Policy Iteration for Budgeted Semantic Video Segmentation