High Quality Monocular Depth Estimation via Transfer Learning

Published in

arXiv.org(2018)

External Links:

Generate Graph

TL;DR

A convolutional neural network for computing a high-resolution depth map given a single RGB image with the help of transfer learning, which outperforms state-of-the-art on two datasets and also produces qualitatively better results that capture object boundaries more faithfully.

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.

Authors

Peter Wonka

11 papers

Ibraheem Alhashim

1 papers

References43 items

Stabilized real-time face tracking via a learned dynamic rigidity prior

Benchmark Analysis of Representative Deep Neural Network Architectures

Detail Preserving Depth Estimation from a Single Image Using Attention Guided Networks

DeepTAM: Deep Tracking and Mapping

Instant 3D photography

High Quality Monocular Depth Estimation via Transfer Learning

Published in

arXiv.org(2018)

External Links:

Generate Graph

TL;DR

Abstract

Authors

Peter Wonka

11 papers

Ibraheem Alhashim

1 papers

References43 items

Stabilized real-time face tracking via a learned dynamic rigidity prior

Benchmark Analysis of Representative Deep Neural Network Architectures

Detail Preserving Depth Estimation from a Single Image Using Attention Guided Networks

DeepTAM: Deep Tracking and Mapping

Instant 3D photography

Digging Into Self-Supervised Monocular Depth Estimation

Deep Ordinal Regression Network for Monocular Depth Estimation

AutoAugment: Learning Augmentation Policies from Data

Taskonomy: Disentangling Task Transfer Learning

DeepMVS: Learning Multi-view Stereopsis

Structured Attention Guided Convolutional Neural Fields for Monocular Depth Estimation

Noise2Noise: Learning Image Restoration without Clean Data

Unsupervised Learning of Depth and Ego-Motion from Video

Multi-scale Continuous CRFs as Sequential Deep Networks for Monocular Depth Estimation

Semi-Supervised Deep Learning for Monocular Depth Map Prediction

DeMoN: Depth and Motion Network for Learning Monocular Stereo

FuseNet: Incorporating Depth into Semantic Segmentation via Fusion-Based CNN Architecture

Unsupervised Monocular Depth Estimation with Left-Right Consistency

Densely Connected Convolutional Networks

Deeper Depth Prediction with Fully Convolutional Residual Networks

Depth and surface normal estimation from monocular images using regression on deep features and hierarchical CRFs

U-Net: Convolutional Networks for Biomedical Image Segmentation

FlowNet: Learning Optical Flow with Convolutional Networks

Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift

Adam: A Method for Stochastic Optimization

Depth Map Prediction from a Single Image using a Multi-Scale Deep Network

Vision meets robotics: The KITTI dataset

ImageNet classification with deep convolutional neural networks

Indoor Segmentation and Support Inference from RGBD Images

Depth-assisted Real-time 3D Object Detection for Augmented Reality

Understanding the difficulty of training deep feedforward neural networks

ImageNet: A large-scale hierarchical image database

Active refocusing of images and videos

Colorization using optimization

Image quality assessment: from error visibility to structural similarity

Epic Games, Inc. Marketplace -UE4 Marketplace

Deeplens: Shallow depth of field from a single image

The Devil is in the Decoders

Tensor-Flow: Large-scale machine learning on heterogeneous systems

Tensor- Method δ1 ↑ δ2 ↑ δ3 ↑ rel↓ rms↓ log10 ↓ mSSIM↑ Laina et al

Rectifier Nonlinearities Improve Neural Network Acoustic Models

A 3x3 isotropic gradient operator for image processing

Visual comparison of estimated depth maps on the Unreal-1k dataset: input RGB images, ground truth depth maps, results using Laina et al. [23], our estimated depth maps

Field of Study

Computer Science

Journal Information

Name

ArXiv

Volume

abs/2005.00687

Venue Information

Name

arXiv.org

Type

URL

https://arxiv.org

Alternate Names

ArXiv