Automatic detection and quantification of floating marine macro-litter in aerial images: Introducing a novel deep learning approach connected to a web application in R.

Asunción Borrell,Toni Monleón-Getino,Luis Cardona,Pere López-Brosa,Ricardo Borja-Robalino,Morgana Vighi,Alex Aguilar,Odei Garcia-Garin

doi:10.1016/j.envpol.2021.116490

Abstract

The threats posed by floating marine macro-litter (FMML) of anthropogenic origin to the marine fauna, and marine ecosystems in general, are universally recognized. Dedicated monitoring programmes and mitigation measures are in place to address this issue worldwide, with the increasing support of new technologies and the automation of analytical processes. In the current study, we developed algorithms capable of detecting and quantifying FMML in aerial images, and a web-oriented application that allows users to identify FMML within images of the sea surface. The proposed algorithm is based on a deep learning approach that uses convolutional neural networks (CNNs) capable of learning from unstructured or unlabelled data. The CNN-based deep learning model was trained and tested using 3723 aerial images (50% containing FMML, 50% without FMML) taken by drones and aircraft over the waters of the NW Mediterranean Sea. The accuracies of image classification (performed using all the images for training and testing the model) and cross-validation (performed using 90% of images for training and 10% for testing) were 0.85 and 0.81, respectively. The Shiny package of R was then used to develop a user-friendly application to identify and quantify FMML within the aerial images. The implementation of this, and similar algorithms, allows streamlining substantially the detection and quantification of FMML, providing support to the monitoring and assessment of this environmental threat. However, the automated monitoring of FMML in the open sea still represents a technological challenge, and further research is needed to improve the accuracy of current algorithms.

Highlights

Marine litter, defined as any persistent, manufactured or processed solid material discarded, disposed of, abandoned, or lost in the marine and coastal environment (UNEP, 2005), is ubiquitous in all marine compartments worldwide (e.g., Arcangeli et al, 2018; Cozar et al, 2014; Suaria et al, 2020)
The aim of the present study was to develop an R (R Core Team, 2020) library based on a deep learning approach, to automatically detect and quantify floating marine macro-litter (FMML) in aerial images of the sea surface taken from drones and aircraft
We applied convolutional neural networks (CNNs)-based deep learning models to detect and quantify FMML in aerial images, we proposed their coupling to the AIImagePred library in R and their implementation on a web-oriented application based on the Shiny package

Summary

Introduction

Marine litter, defined as any persistent, manufactured or processed solid material discarded, disposed of, abandoned, or lost in the marine and coastal environment (UNEP, 2005), is ubiquitous in all marine compartments worldwide (e.g., Arcangeli et al, 2018; Cozar et al, 2014; Suaria et al, 2020). It poses a potential threat to the marine fauna, including invertebrates (e.g., Digka et al, 2018), fish (e.g., Garcia-Garin et al, 2019; 2020d), marine mammals (e.g., De. Stephanis et al, 2013), and turtles (e.g., Schuyler et al, 2014). Floating marine macro-litter (FMML, i.e., objects > 2.5 cm; Galgani et al, 2013; GESAMP, 2019) of anthropogenic origin is harmful, because of its potential to entangle all sort of marine organisms (e.g., fishes, turtles, marine mammals; Deudero & Alomar, 2015), and of being ingested by marine fauna, especially large filterfeeding species (Garcia-Garin et al, 2020c). Monitoring its density and distribution patterns through standardized methodologies (Van Sebille et al, 2020) is highly needed to assess the extent of this environmental threat (GESAMP, 2015; UNEP 2016).

Objectives

Methods

Results

Conclusion