Residual-Guided Learning Representation for Self-Supervised Monocular Depth Estimation

TL;DR

This paper introduces a residual guidance loss that enhances self-supervised monocular depth estimation by transferring discriminative features from auto-encoders, improving stability especially in challenging regions.

Contribution

The proposed residual guidance loss enables depth networks to embed more discriminative features, addressing limitations of existing auto-encoder based feature representations.

Findings

Outperforms existing methods on the KITTI benchmark.

Enhances depth prediction stability in textureless and occluded regions.

Demonstrates orthogonality and complementarity with other state-of-the-art approaches.

Abstract

Photometric consistency loss is one of the representative objective functions commonly used for self-supervised monocular depth estimation. However, this loss often causes unstable depth predictions in textureless or occluded regions due to incorrect guidance. Recent self-supervised learning approaches tackle this issue by utilizing feature representations explicitly learned from auto-encoders, expecting better discriminability than the input image. Despite the use of auto-encoded features, we observe that the method does not embed features as discriminative as auto-encoded features. In this paper, we propose residual guidance loss that enables the depth estimation network to embed the discriminative feature by transferring the discriminability of auto-encoded features. We conducted experiments on the KITTI benchmark and verified our method's superiority and orthogonality on other state-of-the-art methods.