Litcius/Paper detail

Learning What to Learn for Video Object Segmentation

Goutam Bhat, Felix Järemo Lawin, Martin Danelljan, Andreas Robinson, Michael Felsberg, Luc Van Gool, Radu Timofte

2020DiVA (Linkoping University)146 citationsDOIOpen Access PDF

Abstract

Video object segmentation (VOS) is a highly challengingproblem, since the target object is only defined by a first-frame refer-ence mask during inference. The problem of how to capture and utilizethis limited information to accurately segment the target remains a fun-damental research question. We address this by introducing an end-to-end trainable VOS architecture that integrates a differentiable few-shotlearner. Our learner is designed to predict a powerful parametric modelof the target by minimizing a segmentation error in the first frame. Wefurther go beyond the standard few-shot learning paradigm by learningwhat our target model should learn in order to maximize segmentationaccuracy. We perform extensive experiments on standard benchmarks.Our approach sets a new state-of-the-art on the large-scale YouTube-VOS 2018 dataset by achieving an overall score of 81.5, corresponding toa 2.6% relative improvement over the previous best result. The code andmodels are available at https://github.com/visionml/pytracking.

Topics & Concepts

Computer scienceSegmentationArtificial intelligenceFrame (networking)Object (grammar)InferenceComputer visionCode (set theory)Parametric statisticsMachine learningProgramming languageSet (abstract data type)TelecommunicationsMathematicsStatisticsDomain Adaptation and Few-Shot LearningAdvanced Neural Network ApplicationsVisual Attention and Saliency Detection
Learning What to Learn for Video Object Segmentation | Litcius