Abstract
Abstract. Unmanned Aerial Systems (UAS) has been recently used for mapping marine litter on beach-dune environment. Machine learning algorithms have been applied on UAS-derived images and orthophotos for automated marine litter items detection. As sand and vegetation are much predominant on the orthophoto, marine litter items constitute a small set of data, thus a class much less represented on the image scene. This communication aims to analyse the class imbalance issue on orthophotos for automated marine litter items detection. In the used dataset, the percentage of patches containing marine litter is close to 1% of the total amount of patches, hence representing a clear class imbalance issue. This problem has been previously indicated as detrimental for machine learning frameworks. Three different approaches were tested to address this imbalance, namely class weighting, oversampling and classifier thresholding. Oversampling had the best performance with a f1-score of 0.68, while the other methods had f1-score value of 0.56 on average. The results indicate that future works devoted to UAS-based automated marine litter detection should take in consideration the use of the oversampling method, which helped to improve the results of about 7% in the specific case shown in this paper.
Highlights
Coastal marine litter has a negative impact on marine ecosystems (Rochman et al, 2016), marine life (Kühn et al, 2015) coastal communities (Beaumont et al, 2019) and human health (Werner et al, 2016)
The monitoring of marine litter is often based on in-situ visual census surveys (Cheshire et al, 2009), which are expensive in term of human effort and logistically limited (Lavers and Bond, 2017)
The camera gimbal of the Unmanned Aerial Systems (UAS) (DJI Phantom 4) was set to −90° to capture photos perpendicular to the direction of the flight and the ISO, shutter speed and aperture were set to 100, 1/1250 s and f/3.2, respectively. This allowed to generate an orthophoto of 5 mm ground sampling distance (GSD).The orthophoto was generated using 432 UAS images with a resolution of 4864 × 3648 pixels which were acquired at 20 m flying height and overlapped with 80% front and 65% side rates in order to derive the corresponding digital surface model generated by dense image matching
Summary
Coastal marine litter has a negative impact on marine ecosystems (Rochman et al, 2016), marine life (Kühn et al, 2015) coastal communities (Beaumont et al, 2019) and human health (Werner et al, 2016). The monitoring of marine litter is often based on in-situ visual census surveys (Cheshire et al, 2009), which are expensive in term of human effort and logistically limited (Lavers and Bond, 2017). Recent studies have proposed the use of Unmanned Aerial Systems (UAS) for detecting and mapping marine litter pollution on coastal environment (Fallati et al, 2019; Gonçalves et al, 2019). UAS offers a cost-efficient collection of the required high-resolution imagery for the detection of marine meso-litter items (comprised between 2.5 cm and 50 cm) on beaches and dunes. UAS allow a fast coverage of a wider area when compared to in-situ visual census. In spite of this capabilities, manual identification of marine litter items on UAS-derived orthophotos is subjective and time-consuming. As a consequence, automated detection of marine litter items would be preferable, as it would allow a faster and robust image processing
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
