Uav-Based River Plastic Detection With a Multispectral Camera

Abstract

Abstract. Plastic is the third world’s most produced material by industry (after concrete and steel), but people recycle only 9% of plastic that they have used. The other parts are either burned or accumulated in landfills and in the environment, the latter being the cause of many serious consequences, in particular when considering a long-term scenario. A significant part the plastic waste is dispersed in the aquatic environment, having a dramatic impact on the aquatic flora and fauna. This motivated several works aiming at the development of methodologies and automatic or semi-automatic tools for the plastic pollution detection, in order to enable and facilitate its recovery. This paper deals with the problem of plastic waste automatic detection in the fluvial and aquatic environment. The goal is that of exploiting the well-recognized potential of machine learning tools in object detection applications. A machine learning tool, based on random forest classifiers, has been developed to properly detect plastic objects in multi-spectral imagery collected by an unmanned aerial vehicle (UAV). In the developed approach, the outcome is determined by the combination of two random forest classifiers and of an area-based selection criterion. The approach is tested on 154 images collected by a multi-spectral proximity sensor, namely the MAIA-S2 camera, in a fluvial environment, on the Arno river (Italy), where an artificial controlled scenario was created by introducing plastic samples anchored to the ground. The obtained results are quite satisfactory in terms of object detection accuracy and recall (both higher than 98%), while presenting a remarkably lower performance in terms of precision and quality. The overall performance appears also to be dependent on the UAV flight altitude, being worse at higher altitudes, as expected.

Progress in Research on Marine Litter-Related Monitoring Technologies

“…Wolf et al [68] developed APLASTIC-Q, a floating plastic litter identification, classification and quantification system using a machine learning algorithm based on convolutional neural networks, with an accuracy of 83% in identifying water, sand, vegetation and litter on beaches and within rivers in Cambodia. Cortesi et al [69] used an unmanned airborne multispectral camera to monitor floating litter in the Arno River, Italy, and used a stochastic Forest Machine Learning algorithm for litter identification, which showed an accuracy of more than 98% and concluded that the infrared band is helpful in improving the identification accuracy. Maharjan et al [70] processed UAV aerial imagery of the Mekong River tributaries-Huay Mai River and Bangkok Canal for deep learning and concluded that the various YOLO Deep Learning algorithms can all exceed 80% accuracy.…”

Hyperspectral Imaging for Mobile Robot Navigation

“…Environmental sciences use hyperspectral cameras mounted on a drone to detect plastic waste floating on the water surface [ 7 ]. Researchers present a system for the automatic detection of floating plastic waste based on a random forest classifier.…”

“…Spectral angle mapping was used in [ 10 ] as well. Torti et al Cortesi et al [ 7 ] use a random-forest classifier to detect whether a given object floating on a river surface is plastic. Random forest, support vector machine classifiers, and a histogram-based gradient boosting classification tree are used to detect urban material such as asphalt, conglomerate, or sandstone [ 9 ].…”

On the Co-Registration of Asynchronous Multi-Spectral and Thermal Images

“…In particular, different tools and methods have been used over the years to detect and monitor floating plastic objects in fresh and salt water. Some researchers used in-situ visual census for their work (Galgani et al, 2013, Geraeds et al, 2019, but most used images or orthophotos obtained from satellite (Tasseron et al, 2021, Themistocleous et al, 2020, Topouzelis et al, 2020 or Unmanned Aerial Vehicle (UAV) imagery (Cortesi et al, 2022, Jakovljevic et al, 2020, Iordache et al, 2022, Cortesi et al, 2023.…”

“…The use of drones allows not only great flexibility in planning time and location of the surveys but also a wide range of sensors is available, which can improve the overall plastic detection performance. In particular, (Cortesi et al, 2022) suggests that the combination of multi-spectral (in the range of EM wavelength from 433 to 875 nm) and thermal data could be used in order to reduce false positives. * Corresponding author…”