MARIDA: A benchmark for Marine Debris detection from Sentinel-2 remote sensing data

An update for various applications of Artificial Intelligence (AI) for detection and identification of marine environmental pollutions: A bibliometric analysis and systematic review

Marine environmental pollution is one of the growing concerns of humans all over the world. Therefore, managing these marine pollutants has been a crucial matter for scientists in recent decades. Thus, researchers have tried to implement artificial intelligence (AI) to handle marine environmental pollutants. Therefore, in this manuscript, we performed a bibliometric analysis to understand the main applications of AI for managing marine environments. Therefore, we examined both PubMed online database and Google Scholar to find any research articles that discuss the applications of AI in managing marine environmental pollution. Ultimately, we found that AI can detect, locate, and even predict aquatic contaminants like oil fingerprinting, oil spills, oil spill damage, oil slicks, forecasting marine water quality, water quality development, harmful algal blooms, benthic sediment toxicity, as well as detection of marine debris with high accuracy.

Proof of concept for a new sensor to monitor marine litter from space

Worldwide, governments are implementing strategies to combat marine litter. However, their effectiveness is largely unknown because we lack tools to systematically monitor marine litter over broad spatio-temporal scales. Metre-sized aggregations of floating debris generated by sea-surface convergence lines have been reported as a reliable target for detection from satellites. Yet, the usefulness of such ephemeral, scattered aggregations as proxy for sustained, large-scale monitoring of marine litter remains an open question for a dedicated Earth-Observation mission. Here, we track this proxy over a series of 300,000 satellite images of the entire Mediterranean Sea. The proxy is mainly related to recent inputs from land-based litter sources. Despite the limitations of in-orbit technology, satellite detections are sufficient to map hot-spots and capture trends, providing an unprecedented source-to-sink view of the marine litter phenomenon. Torrential rains largely control marine litter inputs, while coastal boundary currents and wind-driven surface sweep arise as key drivers for its distribution over the ocean. Satellite-based monitoring proves to be a real game changer for marine litter research and management. Furthermore, the development of an ad-hoc sensor can lower the minimum detectable concentration by one order of magnitude, ensuring operational monitoring, at least for seasonal-to-interannual variability in the mesoscale.

Large-scale detection of marine debris in coastal areas with Sentinel-2

Detecting and quantifying marine pollution and macroplastics is an increasingly pressing ecological issue that directly impacts ecology and human health. Here, remote sensing can provide reliable estimates of plastic pollution by regularly monitoring and detecting marine debris in coastal areas. In this work, we present a detector for marine debris built on a deep segmentation model that outputs a probability for marine debris at the pixel level. We train this detector with a combination of annotated datasets of marine debris and evaluate it on specifically selected test sites where it is highly probable that plastic pollution is present in the detected marine debris. We integrate data-centric artificial intelligence principles by devising a training strategy with extensive sampling of negative examples and an automated label refinement of coarse hand labels. This yields a deep learning model that achieves higher accuracies on benchmark comparisons than existing detection models trained on previous datasets.

Marine plastic pollution detection and identification by using remote sensing-meta analysis

The persistent plastic litter, originating from different sources and transported from rivers to oceans, has posed serious biological, ecological, and chemical effects on the marine ecosystem, and is considered a global issue. In the past decade, many studies have identified, monitored, and tracked marine plastic debris in coastal and open ocean areas using remote sensing technologies. Compared to traditional surveying methods, high-resolution (spatial and temporal) multispectral or hyperspectral remote sensing data have been substantially used to monitor floating marine macro litter (FMML). In this systematic review, we present an overview of remote sensing data and techniques for detecting FMML, as well as their challenges and opportunities. We reviewed the studies based on different sensors and platforms, spatial and spectral resolution, ground sampling data, plastic detection methods, and accuracy obtained in detecting marine litter. In addition, this study elaborates the usefulness of high-resolution remote sensing data in Visible (VIS), Near-infrared (NIR), and Short-Wave InfraRed (SWIR) range, along with spectral signatures of plastic, in-situ samples, and spectral indices for automatic detection of FMML. Moreover, the Thermal Infrared (TIR), Synthetic aperture radar (SAR), and Light Detection and Ranging (LiDAR) data were introduced and these were demonstrated that could be used as a supplement dataset for the identification and quantification of FMML.

Improving sea surface floating matter identification from Sentinel-2 MSI imagery using optical radiative simulation of neighborhood difference

The reflectance difference (ΔR) between a floating matter pixel and a nearby water reference pixel is a method of atmospheric radiation unmixing. This technique unveils target signals by referencing the background within the horizontal neighborhood. ΔR is effective for removing the mixed-pixel effect and partial atmospheric path radiance. However, other atmospheric interference sources in the difference pixel, including atmospheric extinction and sunglint, need to be clarified. To address these challenges, we combined in situ floating matter endmember spectra for simulation and Sentinel-2 Multispectral Instrument (MSI) sensors for validation. We focused on radiative transfer simulation of horizontal neighborhood and vertical atmospheric column, investigating the bilateral conversion of ΔR between bottom-of-atmosphere (BOA) and top-of-atmosphere (TOA) signals, and clarifying how the atmosphere affects the difference pixel (ΔR) and floating matter identification. Results showed that direct use of TOA ΔR works in discriminating algae from non-algae floating matters under weak sunglint, and is a suitable candidate for no bother with atmospheric correction, least uncertain, and wider coverage. And then, sunglint interference is also inevitable, whether serious or not.

High-precision density mapping of marine debris and floating plastics via satellite imagery

The last couple of years has been ground-breaking for marine pollution monitoring purposes. It has been suggested that combining multi-spectral satellite information and machine learning approaches are effective to monitor plastic pollutants in the ocean environment. Recent research has made theoretical progress in identifying marine debris and suspected plastic (MD&SP) through machine learning whereas no study has fully explored the application of these methods for mapping and monitoring marine debris density. Therefore, this article consists of three main components: (1) the development and validation of a supervised machine learning marine debris detection model, (2) to map the MD&SP density into an automated tool called MAP-Mapper and finally (3) evaluation of the entire system for out-of-distribution (OOD) test locations. Developed MAP-Mapper architectures provide users with options to achieve high precision (abbv. -HP) or optimum precision-recall (abbv. -Opt) values in terms of training/test dataset. Our MAP-Mapper-HP model greatly increases the MD&SP detection precision to 95%, while the MAP-Mapper-Opt achieves 87-88% precision-recall pair. To efficiently measure density mapping findings at OOD test locations, we propose the Marine Debris Map (MDM) index, which combines the average probability of a pixel belonging to the MD&SP class and the number of detections in a given time frame. The high MDM findings of the proposed approach are found to be consistent with existing marine litter and plastic pollution areas, and these are presented with available evidence citing literature and field studies.

Analysis and forecasting of national marine litter based on coastal data in South Korea from 2009 to 2021

In this study, statistical analysis and forecasting were performed using coastal litter data of Korea. The analysis indicated that rope and vinyl accounted for the highest proportion of coastal litter items. The statistical analysis of the national coastal litter trends revealed that the greatest concentration of litter was observed during summer months (June-August). To predict the amount of coastal litter per meter, recurrent neural network (RNN)-based models were used. Neural basis expansion analysis for interpretable time series forecasting (N-BEATS) and neural hierarchical interpolation for time series forecasting (N-HiTS), an improved model of N-BEATS recently announced, were used for comparison with RNN-based models. When predictive performance and trend followability were evaluated, overall N-BEATS and N-HiTS outperformed RNN-based models. Furthermore, we found that average of N-BEATS and N-HiTS models yielded better results than using one model.

Remote detection of marine debris using Sentinel-2 imagery: A cautious note on spectral interpretations

Remote detection of marine debris (also called marine litter) has received increased attention in the past decade, with the Multispectral Instruments (MSI) onboard the Sentinel-2A and Sentinel-2B satellites being the most used sensors. However, because of their mixed band resolutions and small sub-pixel coverage of debris within a pixel (e.g., <10 %), caution is required when interpreting the spectral shapes of MSI pixels. Otherwise, the spectrally distorted shapes may be misused as spectral endmembers (signatures) or interpreted as from certain types of floating matters. Here, using simulations and MSI data, I show the origin of the spectral distortions and emphasize why both pixel averaging and pixel subtraction are critical in algorithm design and spectral interpretation for the purpose of remote detection of marine debris using Sentinel-2 MSI sensors.

Using a UAV Thermal Infrared Camera for Monitoring Floating Marine Plastic Litter

In recent years, the remote sensing of marine plastic litter has been rapidly evolving and the technology is most advanced in the visible (VIS), near-infrared (NIR), and short-wave infrared (SWIR) wavelengths. It has become clear that sensing using VIS-SWIR bands, based on the surface reflectance of sunlight, would benefit from complementary measurements using different technologies. Thermal infrared (TIR) sensing shows potential as a novel method for monitoring macro plastic litter floating on the water surface, as the physics behind surface-leaving TIR is different. We assessed a thermal radiance model for floating plastic litter using a small UAV-grade FLIR Vue Pro R 640 thermal camera by flying it over controlled floating plastic litter targets during the day and night and in different seasons. Experiments in the laboratory supported the field measurements. We investigated the effects of environmental conditions, such as temperatures, light intensity, the presence of clouds, and biofouling. TIR sensing could complement observations from VIS, NIR, and SWIR in several valuable ways. For example, TIR sensing could be used for monitoring during the night, to detect plastics invisible to VIS-SWIR, to discriminate whitecaps from marine litter, and to detect litter pollution over clear, shallow waters. In this study, we have shown the previously unconfirmed potential of using TIR sensing for monitoring floating plastic litter.