Towards Better Data Exploitation in Self-Supervised Monocular Depth Estimation

TL;DR

This paper introduces novel data augmentation and a detail-enhanced network architecture to improve self-supervised monocular depth estimation, achieving state-of-the-art results and better generalization across datasets.

Contribution

It proposes Resizing-Cropping and Splitting-Permuting augmentations combined with self-distillation and a detail-enhanced DepthNet for improved depth estimation.

Findings

Achieves state-of-the-art performance on KITTI benchmark.

Demonstrates superior generalization to Make3D and NYUv2 datasets.

Enhances depth map detail restoration with the new network design.

Abstract

Depth estimation plays an important role in the robotic perception system. Self-supervised monocular paradigm has gained significant attention since it can free training from the reliance on depth annotations. Despite recent advancements, existing self-supervised methods still underutilize the available training data, limiting their generalization ability. In this paper, we take two data augmentation techniques, namely Resizing-Cropping and Splitting-Permuting, to fully exploit the potential of training datasets. Specifically, the original image and the generated two augmented images are fed into the training pipeline simultaneously and we leverage them to conduct self-distillation. Additionally, we introduce the detail-enhanced DepthNet with an extra full-scale branch in the encoder and a grid decoder to enhance the restoration of fine details in depth maps. Experimental results demonstrate our method can achieve state-of-the-art performance on the KITTI benchmark, with both raw ground truth and improved ground truth. Moreover, our models also show superior generalization performance when transferring to Make3D and NYUv2 datasets. Our codes are available at https://github.com/Sauf4896/BDEdepth.