High Quality Monocular Depth Estimation via Transfer Learning

TL;DR

This paper introduces a transfer learning-based convolutional neural network that produces high-resolution, detailed depth maps from single RGB images, outperforming existing methods in accuracy and boundary fidelity.

Contribution

It presents a novel approach leveraging transfer learning and a simple decoder to achieve high-quality depth estimation with fewer parameters and training iterations.

Findings

Outperforms state-of-the-art on two datasets

Produces more detailed and boundary-faithful depth maps

Uses fewer parameters and training iterations

Abstract

Accurate depth estimation from images is a fundamental task in many applications including scene understanding and reconstruction. Existing solutions for depth estimation often produce blurry approximations of low resolution. This paper presents a convolutional neural network for computing a high-resolution depth map given a single RGB image with the help of transfer learning. Following a standard encoder-decoder architecture, we leverage features extracted using high performing pre-trained networks when initializing our encoder along with augmentation and training strategies that lead to more accurate results. We show how, even for a very simple decoder, our method is able to achieve detailed high-resolution depth maps. Our network, with fewer parameters and training iterations, outperforms state-of-the-art on two datasets and also produces qualitatively better results that capture object boundaries more faithfully. Code and corresponding pre-trained weights are made publicly available.