Light-Weight Learning-Based Depth Estimation From A Single Image

Authored by Zekarias Negese, Hartmut Bauermeister, Michael Moeller, Emanuele Rodolà, Zorah Lähner
Published in ATHENA Research Book 2022

Teaser Image

Abstract

Estimating the depth of a scene from a single image is impossible due to scale ambiguity, but recent deep learning approaches have demonstrated to produce faithful estimates anyways by learning the typical scale of objects from implicit clues in the scene. In this paper, we present a new encoder-decoder architecture for single image depth reconstruction that yields state-of-the-art results with considerably less parameters than competitors. Our architecture incorporates a modified inception module tailored to our application which allows us to use a relatively lightweight architecture with fewer learnable parameters than state-of-the-art single image depth reconstruction networks. We train and evaluate on the known NYU RGB-D benchmark dataset, and show generalization of our method by evaluating the pretrained NYU model on the iBims dataset. Our results are in the same order of magnitude as state-of-the-art competitors, even though we trained on a different resolution.

Resources

[pdf]

Bibtex

    @inproceedings{ negese2022lightweight, 
    		author 	= { Zekarias Negese and Hartmut Bauermeister and Michael Moeller and Emanuele Rodolà and Zorah Lähner },
        	title 	= { Light-Weight Learning-Based Depth Estimation From A Single Image },
       		booktitle = { ATHENA Research Book },
        	year 	= { 2022 },
    	}