The Background Also Matters: Background-Aware Motion-Guided Objects Discovery

TL;DR

This paper introduces BMOD, a background-aware method for object discovery in videos that improves segmentation accuracy by explicitly modeling background and foreground, addressing over-segmentation issues in previous motion-guided approaches.

Contribution

The paper proposes a novel background-aware framework that jointly learns object discovery and background separation, significantly enhancing performance over existing methods.

Findings

Improved object discovery accuracy on synthetic and real datasets.

Effective separation of foreground objects from background in unsupervised settings.

Establishment of a strong baseline for object/non-object segmentation.

Abstract

Recent works have shown that objects discovery can largely benefit from the inherent motion information in video data. However, these methods lack a proper background processing, resulting in an over-segmentation of the non-object regions into random segments. This is a critical limitation given the unsupervised setting, where object segments and noise are not distinguishable. To address this limitation we propose BMOD, a Background-aware Motion-guided Objects Discovery method. Concretely, we leverage masks of moving objects extracted from optical flow and design a learning mechanism to extend them to the true foreground composed of both moving and static objects. The background, a complementary concept of the learned foreground class, is then isolated in the object discovery process. This enables a joint learning of the objects discovery task and the object/non-object separation. The conducted experiments on synthetic and real-world datasets show that integrating our background handling with various cutting-edge methods brings each time a considerable improvement. Specifically, we improve the objects discovery performance with a large margin, while establishing a strong baseline for object/non-object separation.