Depth Estimation on Underwater Omni-directional Images Using a Deep Neural Network
This work addresses depth estimation for underwater imaging, which is important for marine robotics and exploration, but it is incremental as it adapts existing methods to a new domain with synthetic data.
The authors tackled depth estimation for underwater perspective and omni-directional images by adapting a Fully Convolutional Residual Neural Network (FCRN) from in-air to underwater settings, using synthetic data due to lack of real datasets, and demonstrated effectiveness through qualitative and quantitative comparisons with ground truth.
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.