Video Object Segmentation with Adaptive Feature Bank and Uncertain-Region Refinement

TL;DR

This paper introduces an adaptive feature bank update scheme and an uncertain-region refinement method to improve semi-supervised video object segmentation, achieving state-of-the-art results on public benchmarks.

Contribution

It presents a novel adaptive feature bank update mechanism and a confidence-based refinement module for more accurate and efficient video object segmentation.

Findings

Outperforms existing state-of-the-art methods on public benchmarks.

Effective organization of feature banks improves segmentation accuracy.

Refinement in uncertain regions enhances overall segmentation quality.

Abstract

We propose a new matching-based framework for semi-supervised video object segmentation (VOS). Recently, state-of-the-art VOS performance has been achieved by matching-based algorithms, in which feature banks are created to store features for region matching and classification. However, how to effectively organize information in the continuously growing feature bank remains under-explored, and this leads to inefficient design of the bank. We introduce an adaptive feature bank update scheme to dynamically absorb new features and discard obsolete features. We also design a new confidence loss and a fine-grained segmentation module to enhance the segmentation accuracy in uncertain regions. On public benchmarks, our algorithm outperforms existing state-of-the-arts.