Source-free Depth for Object Pop-out

TL;DR

This paper introduces a method that adapts depth inference models for 3D object segmentation using a 'pop-out' prior, enabling better object localization and segmentation without source data, demonstrated across multiple datasets.

Contribution

It proposes a novel adaptation approach that leverages inferred depth maps and a contact surface prior for 3D object segmentation, improving generalization and efficiency.

Findings

Enhanced segmentation performance on camouflaged and salient objects.

Improved generalization across diverse datasets.

Efficient use of only depth models without source data.

Abstract

Depth cues are known to be useful for visual perception. However, direct measurement of depth is often impracticable. Fortunately, though, modern learning-based methods offer promising depth maps by inference in the wild. In this work, we adapt such depth inference models for object segmentation using the objects' "pop-out" prior in 3D. The "pop-out" is a simple composition prior that assumes objects reside on the background surface. Such compositional prior allows us to reason about objects in the 3D space. More specifically, we adapt the inferred depth maps such that objects can be localized using only 3D information. Such separation, however, requires knowledge about contact surface which we learn using the weak supervision of the segmentation mask. Our intermediate representation of contact surface, and thereby reasoning about objects purely in 3D, allows us to better transfer the depth knowledge into semantics. The proposed adaptation method uses only the depth model without needing the source data used for training, making the learning process efficient and practical. Our experiments on eight datasets of two challenging tasks, namely camouflaged object detection and salient object detection, consistently demonstrate the benefit of our method in terms of both performance and generalizability.