A Stereo‐Matching Technique for Recovering 3D Information from Underwater Inspection Imagery

et al. 2018

JMSE

Self Cite

Recent breakthroughs in the computer vision community have led to the emergence of efficient deep learning techniques for end-to-end segmentation of natural scenes. Underwater imaging stands to gain from these advances, however, deep learning methods require large annotated datasets for model training and these are typically unavailable for underwater imaging applications. This paper proposes the use of photorealistic synthetic imagery for training deep models that can be applied to interpret real-world underwater imagery. To demonstrate this concept, we look at the specific problem of biofouling detection on marine structures. A contemporary deep encoder-decoder network, termed SegNet, is trained using 2500 annotated synthetic images of size 960 × 540 pixels. The images were rendered in a virtual underwater environment under a wide variety of conditions and feature biofouling of various size, shape, and colour. Each rendered image has a corresponding ground truth per-pixel label map. Once trained on the synthetic imagery, SegNet is applied to segment new real-world images. The initial segmentation is refined using an iterative support vector machine (SVM) based post-processing algorithm. The proposed approach achieves a mean Intersection over Union (IoU) of 87% and a mean accuracy of 94% when tested on 32 frames extracted from two distinct real-world subsea inspection videos. Inference takes several seconds for a typical image.

Section: Introductionmentioning

confidence: 99%

Semantic Segmentation of Underwater Imagery Using Deep Networks Trained on Synthetic Imagery

O’Byrne

Pakrashi

et al. 2018

JMSE

Self Cite

“…Indeed, our data came from videotape analysis and roughness is hard to measure when using pointing algorithm to extract information from videotape frames. However, it could be done using advanced image processing algorithm such as the work of O'Byrne et al [13], which was not available at the time of the study. Concerning density, expressed in kilograms per meter cube, it was hypothesised that density is homogeneous along the mooring line.…”

Section: An a Priori Spatial Distribution Of Bio-colonisationmentioning

confidence: 99%

Model of Bio-Colonisation on Mooring Lines: Updating Strategy Based on a Static Qualifying Sea State for Floating Wind Turbines

Decurey

Barillé³

et al. 2020

JMSE

Self Cite

Bio-colonisation affects the ageing of materials and the behaviour of offshore structures. Mooring systems and umbilicals belong to the family of slender bodies which are components sensitive to bio-colonisation because of a change of dynamic behaviour due to shape, roughness and mass modifications. However, this stochastic process in time and space is hard to predict. The purpose is then twofold: first, to provide a stochastic spatial model of the bio-colonisation on a mooring line; second, to show that in some defined environmental conditions, such as low wave height, low wind and current velocities, the monitoring of mooring lines tension can help to assess and reduce uncertainty on this model. Therefore, a comprehensive stochastic modelling based on mussels colonisation was carried out using on-site videotapes, experimental campaigns and expert knowledge. We studied the efficiency of a virtual sensing network using this model and a conditional entropy metric. It is first shown that the spatial model fits well with experimental data, and second that a denser medium accuracy sensor network is to be preferred to a single high accuracy fairlead sensor to reduce the uncertainty on the model parameters. It is then worth updating bio-colonisation on mooring lines during the life-time of a floating wind turbine.

“…The latter is a model error for modeling the uncertainty when quantifying the real effect of a randomly distributed roughness around the component. Note that intensive developments on underwater image processing are emerging [54][55][56], enabling one to envisage progress in on-site measurements. Recent works investigate the relationship between non-homogenous roughness and loading [57,58].…”

Section: Stochastic Modeling Of Marine Growth and Hydrodynamic Paramementioning

confidence: 99%

Stochastic Modeling of Forces on Jacket-Type Offshore Structures Colonized by Marine Growth

Ameryoun

Barillé

et al. 2019

JMSE

Self Cite

The present paper deals with the stochastic modeling of bio-colonization for the computation of stochastic hydrodynamic loading on jacket-type offshore structures. It relies on a multidisciplinary study gathering biological and physical research fields that accounts for uncertainties at all the levels. Indeed, bio-colonization of offshore structures is a complex phenomenon with two major but distinct domains: (i) marine biology, whose processes are modeled with biomathematics methods, and (ii) hydrodynamic processes. This paper aims to connect these two domains. It proposes a stochastic model for the marine organism’s growth and then continues with transfers for the assessment of drag coefficient and forces probability density functions that account for marine growth evolution. A case study relies on the characteristics (growth and shape) of the blue mussel (Mytilus edulis) in the northeastern Atlantic.