Depth Estimation on Underwater Omni-directional Images Using a Deep Neural Network

TL;DR

This paper presents a deep neural network approach for estimating depth in underwater omni-directional images, utilizing synthetic data and spherical convolutional networks to improve accuracy in challenging underwater environments.

Contribution

The work introduces a spherical Fully Convolutional Residual Neural Network for underwater omni-directional depth estimation, adapting perspective models to spherical images and synthesizing data for training.

Findings

Effective depth estimation demonstrated on synthetic underwater images.

Spherical FCRN outperforms perspective models in accuracy.

Qualitative and quantitative results validate the approach.

Abstract

In this work, we exploit a depth estimation Fully Convolutional Residual Neural Network (FCRN) for in-air perspective images to estimate the depth of underwater perspective and omni-directional images. We train one conventional and one spherical FCRN for underwater perspective and omni-directional images, respectively. The spherical FCRN is derived from the perspective FCRN via a spherical longitude-latitude mapping. For that, the omni-directional camera is modeled as a sphere, while images captured by it are displayed in the longitude-latitude form. Due to the lack of underwater datasets, we synthesize images in both data-driven and theoretical ways, which are used in training and testing. Finally, experiments are conducted on these synthetic images and results are displayed in both qualitative and quantitative way. The comparison between ground truth and the estimated depth map indicates the effectiveness of our method.