Deeper Depth Prediction with Fully Convolutional Residual Networks

TLDR

Depth estimation from a single RGB image is the problem addressed. The authors propose a fully convolutional residual network to model the mapping from monocular images to depth maps. The model is a single end‑to‑end fully convolutional residual network that learns efficient up‑sampling, uses a reverse Huber loss, and requires no post‑processing. The network runs in real time, uses fewer parameters and training data than state‑of‑the‑art methods, outperforms them on depth estimation, and its code is publicly available.

Abstract

This paper addresses the problem of estimating the depth map of a scene given a single RGB image. We propose a fully convolutional architecture, encompassing residual learning, to model the ambiguous mapping between monocular images and depth maps. In order to improve the output resolution, we present a novel way to efficiently learn feature map up-sampling within the network. For optimization, we introduce the reverse Huber loss that is particularly suited for the task at hand and driven by the value distributions commonly present in depth maps. Our model is composed of a single architecture that is trained end-to-end and does not rely on post-processing techniques, such as CRFs or other additional refinement steps. As a result, it runs in real-time on images or videos. In the evaluation, we show that the proposed model contains fewer parameters and requires fewer training data than the current state of the art, while outperforming all approaches on depth estimation. Code and models are publicly available.