One-Shot Video Object Segmentation

TL;DR

This paper introduces OSVOS, a one-shot video object segmentation method using a fully-convolutional neural network that effectively segments objects in videos with high accuracy and temporal coherence, based on minimal initial annotation.

Contribution

The paper presents a novel one-shot segmentation approach that transfers semantic knowledge from ImageNet to individual object segmentation in videos, achieving state-of-the-art results.

Findings

OSVOS achieves 79.8% accuracy, surpassing previous methods.

The method is fast and produces temporally coherent segmentation.

It effectively transfers learned features from generic to specific object segmentation.

Abstract

This paper tackles the task of semi-supervised video object segmentation, i.e., the separation of an object from the background in a video, given the mask of the first frame. We present One-Shot Video Object Segmentation (OSVOS), based on a fully-convolutional neural network architecture that is able to successively transfer generic semantic information, learned on ImageNet, to the task of foreground segmentation, and finally to learning the appearance of a single annotated object of the test sequence (hence one-shot). Although all frames are processed independently, the results are temporally coherent and stable. We perform experiments on two annotated video segmentation databases, which show that OSVOS is fast and improves the state of the art by a significant margin (79.8% vs 68.0%).