Large quantities of mismanaged plastic waste are polluting and threatening the health of the blue planet. As such, vast amounts of this plastic waste found in the oceans originates from land. It finds its way to the open ocean through rivers, waterways and estuarine systems. Here we present a novel machine learning algorithm based on convolutional neural networks (CNNs) that is capable of detecting and quantifying floating and washed ashore plastic litter. The aquatic plastic litter detection, classification and quantification system (APLASTIC-Q) was developed and trained using very high geo-spatial resolution imagery (∼5 pixels cm−1 = 0.002 m pixel−1) captured from aerial surveys in Cambodia. APLASTIC-Q was made up of two machine learning components (i) plastic litter detector (PLD-CNN) and (ii) plastic litter quantifier (PLQ-CNN). PLD-CNN managed to categorize targets as water, sand, vegetation and plastic litter with an 83% accuracy. It also provided a qualitative count of litter as low or high based on a thresholding approach. PLQ-CNN further distinguished and enumerated the litter items in each of the classes defined as water bottles, Styrofoam, canisters, cartons, bowls, shoes, polystyrene packaging, cups, textile, carry bags small or large. The types and amounts of plastic litter provide benchmark information that is urgently needed for decision-making by policymakers, citizens and other public and private stakeholders. Quasi-quantification was based on automated counts of items present in the imagery with caveats of underlying object in case of aggregated litter. Our scientific evidence-based machine learning algorithm has the prospects of complementing net trawl surveys, field campaigns and clean-up activities for improved quantification of plastic litter. APLASTIC-Q is a smart algorithm that is easy to adapt for fast and automated detection as well as quantification of floating or washed ashore plastic litter from aerial, high-altitude pseudo satellites and space missions.

No abstract

<p> Plastic pollution has a big impact on living organisms. At the same time, plastics are everywhere in our daily life. For example, plastic is used in packaging, construction of buildings, cars, electronics, agriculture and many other fields. In fact, plastic production has been increasing rapidly since the 1950s. However, plastic waste management strategies have not adapted accordingly to these rising amounts, which end up in the blue and green planet. Unfortunately, for developing nations it is even more complicated and strategies are still developing. Here we investigate the possibilities of plastic waste detection in Cambodia focusing on cities, rivers and coastal areas. Very fine geo-spatial resolution Red-Green-Blue (RGB) drone imagery was captured over regions of interest in Phnom Penh, Sihanoukville and Siem Reap. To this date, techniques of detecting plastic litter are based on RGB imagery analyses, generating descriptors such as colour, shape, size and form. However, we believe by adding infrared wavebands additional descriptors, such as polymer composition or type can be retrieved for improved classification of plastic litter. Furthermore, remote sensing technologies will be merged with object-based deep learning methodologies to enhance identification of plastic waste items, thus creating a robust learning system. Due to the size and complexity of this problem, automated detection, tracking, characterization and quantification of plastic pollution is a key aspect to improve waste management strategies. We therefore explore multispectral band combinations relevant to the detection of plastic waste and operational approaches in imagery processing. This work will contribute towards algorithm development for analysis of video datasets enhancing future near real-time detection of plastic litter. Eventually, this scientific evidence-based tool can be utilized by stakeholders, policymakers and citizens.</p>

<p>Plastic Litter (PL) has become more ubiquitous in the last decades posing socio-economic as well as health problems for the blue and green economy. However, to date PL monitoring strategies have been based on field sampling by citizens and scientists during recreational, sporting, scientific and clean-up campaigns. To this end, remote sensing technologies combined with artificial intelligence (AI) have gained rising interest as a potential source of complementary scientific evidence-based information with the capabilities to (i) detect, (ii) track, (iii) characterise and (iv) quantify PL. Within the smart algorithms, convoluted and recurrent neural networks ingest vast multi to hyperspectral images from smartphones, unmanned aerial systems, fixed observatories, high-altitude pseudo-satellites and space stations. Detection would involve the application of object recognition algorithms to true colour Red-Green-Blue (RGB) composite images. Typical essential descriptors that are derived from RGB images include apparent colour, shape, type and dimensions of PL. In addition to object recognition algorithms supported by visual inspection, AI is also used to classify and estimate counts of PL in captured imagery. Quantification assisted by smart systems have the advantage of uncertainties associated with predictions, a cruial aspect in determing budgets of PL in the natural environment. Hyperspectral data is then utilized to further characterise the polymer composition of PL based on spectral reference libraries of known polymers. Fixed observatories and repeated image capture at regions-of-interest have prospective applications in tracking of PL. Here we present plausible applications of remote detection, tracking and quantification of PL assisted by smart AI algorithms. Smart remote sensing of PL will be integrated in future operational smart observing system with near real-time capabilities to generate user (citizens, stakeholders, policymakers) defined end-products relevant to plastic litter. These tailor-made descriptors will thus contribute towards scientific evidence-based knowledge important in assisting legislature in policy making, awareness campaigns as well as evaluating the efficacy of mitigation strategies for plastic litter. Essential descriptors proposed need to include geolocations, quantities, size distributions, shape/form, apparent colour and polymer composition of PL.</p>