Self-Supervised Video Object Segmentation by Motion-Aware Mask Propagation

TL;DR

This paper introduces MAMP, a self-supervised method for video object segmentation that uses motion-aware mask propagation and frame reconstruction, achieving state-of-the-art results without annotations.

Contribution

The paper presents a novel self-supervised approach called MAMP that effectively handles fast motion and long-term matching in video segmentation without requiring labeled data.

Findings

MAMP outperforms existing self-supervised methods on DAVIS-2017 and YouTube-VOS datasets.

MAMP achieves performance comparable to supervised methods.

MAMP demonstrates strong generalization to unseen categories.

Abstract

We propose a self-supervised spatio-temporal matching method, coined Motion-Aware Mask Propagation (MAMP), for video object segmentation. MAMP leverages the frame reconstruction task for training without the need for annotations. During inference, MAMP extracts high-resolution features from each frame to build a memory bank from the features as well as the predicted masks of selected past frames. MAMP then propagates the masks from the memory bank to subsequent frames according to our proposed motion-aware spatio-temporal matching module to handle fast motion and long-term matching scenarios. Evaluation on DAVIS-2017 and YouTube-VOS datasets show that MAMP achieves state-of-the-art performance with stronger generalization ability compared to existing self-supervised methods, i.e., 4.2% higher mean J&F on DAVIS-2017 and 4.85% higher mean J&F on the unseen categories of YouTube-VOS than the nearest competitor. Moreover, MAMP performs at par with many supervised video object segmentation methods. Our code is available at: https://github.com/bo-miao/MAMP.