Multi-Granularity Video Object Segmentation

TL;DR

This paper introduces MUG-VOS, a large-scale multi-granularity video segmentation dataset for both salient and non-salient objects, and proposes a memory-based mask propagation model that outperforms existing methods.

Contribution

The paper creates a new multi-granularity video segmentation dataset and develops a novel memory-based mask propagation model for improved segmentation performance.

Findings

MUG-VOS dataset includes diverse granularities of object masks.

The proposed MMPM model achieves state-of-the-art results.

Automatic and human annotations ensure dataset reliability.

Abstract

Current benchmarks for video segmentation are limited to annotating only salient objects (i.e., foreground instances). Despite their impressive architectural designs, previous works trained on these benchmarks have struggled to adapt to real-world scenarios. Thus, developing a new video segmentation dataset aimed at tracking multi-granularity segmentation target in the video scene is necessary. In this work, we aim to generate multi-granularity video segmentation dataset that is annotated for both salient and non-salient masks. To achieve this, we propose a large-scale, densely annotated multi-granularity video object segmentation (MUG-VOS) dataset that includes various types and granularities of mask annotations. We automatically collected a training set that assists in tracking both salient and non-salient objects, and we also curated a human-annotated test set for reliable evaluation. In addition, we present memory-based mask propagation model (MMPM), trained and evaluated on MUG-VOS dataset, which leads to the best performance among the existing video object segmentation methods and Segment SAM-based video segmentation methods. Project page is available at https://cvlab-kaist.github.io/MUG-VOS.