A novel deep learning instance segmentation model for automated marine oil spill detection

Yekeen, Shamsudeen Temitope; Balogun, Abdul‐Lateef; Yusof, Khamaruzaman B. Wan

doi:10.1016/j.isprsjprs.2020.07.011

Cited by 102 publications

(40 citation statements)

References 51 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Recent deep-learning-based methods [9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27][28] that used CNN structures for both automated feature extraction, as well as classification of SAR images have relied on the use of patches to reduce the background concentration in the tested images. Pre-trained models, such as ResNet 101, VGG-16, and GAN networks as in [10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27][28] or multi-level CNN networks as in [9], were introduced to classify patches with modest performance (i.e., precision and Dice score). In [23], authors obtained improved results via introducing the use of VGG-16 and dark batch generation algorithm on spaceborne SAR images.…”

Section: Discussionmentioning

confidence: 99%

“…Yekeen et al [24,25] introduced the use of mask-region-based CNN to distinguish between ships, oil spills, and look-alikes where pre-trained ResNet 101 and feature pyramid network were used for feature extraction, regional proposal network was deployed for the region of interest extraction, and the mask-region-based CNN was used for semantic segmentation. The proposed model introduced a classification accuracy of 96%, and 92% for oil spills and look-alikes, respectively.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

A Deep-Learning Framework for the Detection of Oil Spills from SAR Data

Shaban

Salim

Khalifeh

et al. 2021

Sensors

View full text Add to dashboard Cite

Oil leaks onto water surfaces from big tankers, ships, and pipeline cracks cause considerable damage and harm to the marine environment. Synthetic Aperture Radar (SAR) images provide an approximate representation for target scenes, including sea and land surfaces, ships, oil spills, and look-alikes. Detection and segmentation of oil spills from SAR images are crucial to aid in leak cleanups and protecting the environment. This paper introduces a two-stage deep-learning framework for the identification of oil spill occurrences based on a highly unbalanced dataset. The first stage classifies patches based on the percentage of oil spill pixels using a novel 23-layer Convolutional Neural Network. In contrast, the second stage performs semantic segmentation using a five-stage U-Net structure. The generalized Dice loss is minimized to account for the reduced oil spill representation in the patches. The results of this study are very promising and provide a comparable improved precision and Dice score compared to related work.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

A Deep-Learning Framework for the Detection of Oil Spills from SAR Data

Shaban

Salim

Khalifeh

et al. 2021

Sensors

View full text Add to dashboard Cite

show abstract

“…The Mask-RCNN outperformed their algorithm. Yekeen et al [77] applied Mask-RCNN in oil spill detection using Keras and Tensorflow and ResNet101-FPN backbone in Synthetic-Aperture Radar (SAR) imagery. The authors analyzed precision, recall, specificity, f1, IoU, and overall accuracy, showing promising results.…”

Section: Discussionmentioning

confidence: 99%

“…Instance segmentation has applications in several areas of knowledge: medicine [69,70], biology [71,72], livestock [73,74], agronomy [75,76], among others. However, remote sensing application is still restricted, highlighting its use in the automatic detection of the following targets: marine oil spill [77], building [78,79], vehicle [80], and ship [81].…”

Section: Introductionmentioning

confidence: 99%

Instance Segmentation for Large, Multi-Channel Remote Sensing Imagery Using Mask-RCNN and a Mosaicking Approach

et al. 2020

View full text Add to dashboard Cite

Instance segmentation is the state-of-the-art in object detection, and there are numerous applications in remote sensing data where these algorithms can produce significant results. Nevertheless, one of the main problems is that most algorithms use Red, Green, and Blue (RGB) images, whereas Satellite images often present more channels that can be crucial to improve performance. Therefore, the present work brings three contributions: (a) conversion system from ground truth polygon data into the Creating Common Object in Context (COCO) annotation format; (b) Detectron2 software source code adaptation and application on multi-channel imagery; and (c) large scene image mosaicking. We applied the procedure in a Center Pivot Irrigation System (CPIS) dataset with ground truth produced by the Brazilian National Water Agency (ANA) and Landsat-8 Operational Land Imager (OLI) imagery (7 channels with 30-m resolution). Center pivots are a modern irrigation system technique with massive growth potential in Brazil and other world areas. The round shapes with different textures, colors, and spectral behaviors make it appropriate to use Deep Learning instance segmentation. We trained the model using 512 × 512-pixel sized patches using seven different backbone structures (ResNet50- Feature Pyramid Network (FPN), Resnet50-DC5, ResNet50-C4, Resnet101-FPN, Resnet101-DC5, ResNet101-FPN, and ResNeXt101-FPN). The model evaluation used standard COCO metrics (Average Precision (AP), AP50, AP75, APsmall, APmedium, and AR100). ResNeXt101-FPN had the best results, with a 3% advantage over the second-best model (ResNet101-FPN). We also compared the ResNeXt101-FPN model in the seven-channel and RGB imagery, where the multi-channel model had a 3% advantage, demonstrating great improvement using a larger number of channels. This research is also the first with a mosaicking algorithm using instance segmentation models, where we tested in a 1536 × 1536-pixel image using a non-max suppression sorted by area method. The proposed methodology is innovative and suitable for many other remote sensing problems and medical imagery that often present more channels.

show abstract

“…The first three indicators, specifically Precision (P), Recall (R), and F1 score (F1), were generated based on the True Positive (TP), True Negative (TN), False Positive (FP), and False Negative (FN) [61,70]. Precision shows the ratio of the correctly classified classes that are positive for each class.…”

Section: Accuracy Assessmentmentioning

confidence: 99%

Comparing Methods for Segmenting Supra-Glacial Lakes and Surface Features in the Mount Everest Region of the Himalayas Using Chinese GaoFen-3 SAR Images

Chen

2021

Remote Sensing

View full text Add to dashboard Cite

Glaciers and numerous glacial lakes that are produced by glacier melting are key indicators of climate change. Often overlooked, supra-glacial lakes develop in the melting area in the low-lying part of a glacier and appear to be highly variable in their size, shape, and location. The lifespan of these lakes is thought to be quite transient, since the lakes may be completely filled by water and burst out within several weeks. Changes in supra-glacial lake outlines and other surface features such as supra-glacial rivers and crevasses on the glaciers are useful indicators for the direct monitoring of glacier changes. Synthetic aperture radar (SAR) is not affected by weather and climate, and is an effective tool for study of glaciated areas. The development of the Chinese GaoFen-3 (GF-3) SAR, which has high spatial and temporal resolution and high-precision observation performance, has made it possible to obtain dynamic information about glaciers in more detail. In this paper, the classical Canny operator, the variational B-spline level-set method, and U-Net-based deep-learning model were applied and compared to extract glacial lake outlines and other surface features using different modes and Chinese GF-3 SAR imagery in the Mount Everest Region of the Himalayas. Particularly, the U-Net-based deep-learning method, which was independent of auxiliary data and had a high degree of automation, was used for the first time in this context. The experimental results showed that the U-Net-based deep-learning model worked best in the segmentation of supra-glacial lakes in terms of accuracy (Precision = 98.45% and Recall = 95.82%) and segmentation efficiency, and was good at detecting small, elongated, and ice-covered supra-glacial lakes. We also found that it was useful for accurately identifying the location of supra-glacial streams and ice crevasses on glaciers, and quantifying their width. Finally, based on the time series of the mapping results, the spatial characteristics and temporal evolution of these features over the glaciers were comprehensively analyzed. Overall, this study presents a novel approach to improve the detection accuracy of glacier elements that could be leveraged for dynamic monitoring in future research.

show abstract

A novel deep learning instance segmentation model for automated marine oil spill detection

Cited by 102 publications

References 51 publications

A Deep-Learning Framework for the Detection of Oil Spills from SAR Data

A Deep-Learning Framework for the Detection of Oil Spills from SAR Data

Instance Segmentation for Large, Multi-Channel Remote Sensing Imagery Using Mask-RCNN and a Mosaicking Approach

Comparing Methods for Segmenting Supra-Glacial Lakes and Surface Features in the Mount Everest Region of the Himalayas Using Chinese GaoFen-3 SAR Images

Contact Info

Product

Resources

About