Projects | Computational Imaging Lab

Unsupervised Single Image Underwater Depth Estimation

Accepted for publication at ICIP 2019 [paper] [code]

Honey Gupta and Kaushik Mitra

unsupervised-single-image-underwater-depth-estimation

Figure 1: UW-Net : Our approach is based on indirect depth-map estimation, where we learn the mapping between unpaired RGB-D terrestrial images and arbitrary underwater images to implicitly estimate the required depth-map. Our method is based on the principles of cycle-consistent learning and uses dense-block based auto-encoders as generator networks. Generator G transforms underwater images to above-water color and depth images whereas F converts above-water color and depth to underwater color image. G is constrained to learn depth estimation from underwater image due the L_cyc between above-water y and G(F(y)), thus implicitly estimation depth for real underwater images.

Abstract

Depth estimation from a single underwater image is one of the most challenging problems and is highly ill-posed. Due to the absence of large underwater depth datasets and the difficulty in obtaining ground truth depth-maps, supervised learning techniques such as direct depth regression cannot be used. In this paper, we propose an unsupervised method for depth estimation from a single underwater image taken “in the wild” by using haze as a cue for depth. Our approach is based on indirect depth-map estimation where we learn the mapping between unpaired RGB-D terrestrial images and arbitrary underwater images to implicitly estimate the required depth-map. We propose a method which is based on the principles of cycle-consistent learning and uses dense-block based auto-encoders as generator networks. We evaluate and compare our method both quantitatively and qualitatively on various underwater images with diverse attenuation and scattering conditions and show that our method produces state-of-the-art results for unsupervised depth estimation from a single underwater image.

Figure 2 : Instead of following the usual approach of obtaining depthmap from transmission-map, we propose a novel depth estimation network based on implicit learning. Column 2-3 show results from transmission-map based methods; column 4 shows our results.

Publication

@article{gupta2019unsupervised,
title={Unsupervised Single Image Underwater Depth Estimation},
author={Gupta, Honey and Mitra, Kaushik},
journal={arXiv preprint arXiv:1905.10595},
year={2019} }

Results

Unsupervised single underwater image depth estimation results

UW-Net depth-map estimates for randomly taken underwater images from our collected dataset belonging to different illumination and scattering conditions.

Figure below shows unsupervised single image depth-estimation results for different scenes.