Search for: All records

Underwater image enhancement and turbidity removal (dehazing) is a very challenging problem, not only due to the sheer variety of environments where it is applicable, but also due to the lack of high-resolution, labelled image data. In this paper, we present a novel, two-step deep learning approach for underwater image dehazing and colour correction. In iDehaze, we leverage computer graphics to physically model light propagation in underwater conditions. Specifically, we construct a three-dimensional, photorealistic simulation of underwater environments, and use them to gather a large supervised training dataset. We then train a deep convolutional neural network to remove the haze in these images, then train a second network to transform the colour space of the dehazed images onto a target domain. Experiments demonstrate that our two-step iDehaze method is substantially more effective at producing high-quality underwater images, achieving state-of-the-art performance on multiple datasets. Code, data and benchmarks will be open sourced. 

Sensing and communication technology has been used successfully in various event monitoring applications over the last two decades, especially in places where long-term manual monitoring is infeasible. However, the major applicability of this technology was mostly limited to terrestrial environments. On the other hand, underwater wireless sensor networks (UWSNs) opens a new space for the remote monitoring of underwater species and faunas, along with communicating with underwater vehicles, submarines, and so on. However, as opposed to terrestrial radio communication, underwater environment brings new challenges for reliable communication due to the high conductivity of the aqueous medium which leads to major signal absorption. In this paper, we provide a detailed technical overview of different underwater communication technologies, namely acoustic, magnetic, and visual light, along with their potentials and challenges in submarine environments. Detailed comparison among these technologies have also been laid out along with their pros and cons using real experimental results. 

Optical camera communication is an emerging technology that enables communication using light beams, where information is modulated through optical transmissions from light-emitting diodes (LEDs). This work conducts empirical studies to identify the feasibility and effectiveness of using deep learning models to improve signal reception in camera communication. The key contributions of this work include the investigation of transfer learning and customization of existing models to demodulate the signals transmitted using a single LED by applying the classification models on the camera frames at the receiver. In addition to investigating deep learning methods for demodulating a single VLC transmission, this work evaluates two real-world use-cases for the integration of deep learning in visual multiple-input multiple-output (MIMO), where transmissions from a LED array are decoded on a camera receiver. This paper presents the empirical evaluation of state-of-the-art deep neural network (DNN) architectures that are traditionally used for computer vision applications for camera communication. 

In this paper, we present OpenWaters, a real-time open-source underwater simulation kit for generating photorealistic underwater scenes. OpenWaters supports creation of massive amount of underwater images by emulating diverse real-world conditions. It allows for fine controls over every variable in a simulation instance, including geometry, rendering parameters like ray-traced water caustics, scattering, and ground-truth labels. Using underwater depth (distance between camera and object) estimation as the use-case, we showcase and validate the capabilities of OpenWaters to model underwater scenes that are used to train a deep neural network for depth estimation. Our experimental evaluation demonstrates depth estimation using synthetic underwater images with high accuracy, and feasibility of transfer-learning of features from synthetic to real-world images.