Beach litter monitoring programs play a key role in establishing effective management measures to preserve the ecological, scenic, and economic value of the coastal areas. In this study, an innovative analysis system is proposed for the automatic identification of beach debris (>2.5 cm) on aerial-photogrammetric images acquired by unmanned aerial vehicles (UAV) at different elevations. The workflow is based on a Convolutional Neural Network model designed for object segmentation and object recognition, and here used for instance segmentation tasks. Test cases were conducted along the Adriatic sector of the Apulia region (Italy), where the beaches have a remarkable economic importance, attracting national and international tourists, and ecological values, hosting species of high ecological value and protected areas. The results of the tests carried out in this study allowed defining 10 m as the desirable drone flight above ground. In addition, encouraging results have been obtained on the instance segmentation step, experimenting on real, synthetic and mixed data, produced by using a high-resolution blending technique. A beach litter density of 0.38 items m(-2) (on 0.66 items m(-2)), an F-score of 0.96 and a mAP of 0.67 has been achieved on real data. A novel metric for comparing works at the state-of-the-art (SOTA) in beach litter monitoring is also introduced, named "density-normalized F-score". The proposed methodology represents a benchmark for the definition of a standardize procedure for the indirect evaluation and monitoring of the coastal environmental status. Besides allowing the investigation of large areas with limited human effort, the proposed system enables the evaluation of the beach litter spatial distribution and magnitude, providing useful information for the assessment of tailored beach quality indices.

A novel beach litter analysis system based on UAV images and Convolutional Neural Networks

Scarrica, VM;Aucelli, PPC;Casolaro, A;Fiore, P;Rizzo, A;Staiano, A
2022-01-01

Abstract

Beach litter monitoring programs play a key role in establishing effective management measures to preserve the ecological, scenic, and economic value of the coastal areas. In this study, an innovative analysis system is proposed for the automatic identification of beach debris (>2.5 cm) on aerial-photogrammetric images acquired by unmanned aerial vehicles (UAV) at different elevations. The workflow is based on a Convolutional Neural Network model designed for object segmentation and object recognition, and here used for instance segmentation tasks. Test cases were conducted along the Adriatic sector of the Apulia region (Italy), where the beaches have a remarkable economic importance, attracting national and international tourists, and ecological values, hosting species of high ecological value and protected areas. The results of the tests carried out in this study allowed defining 10 m as the desirable drone flight above ground. In addition, encouraging results have been obtained on the instance segmentation step, experimenting on real, synthetic and mixed data, produced by using a high-resolution blending technique. A beach litter density of 0.38 items m(-2) (on 0.66 items m(-2)), an F-score of 0.96 and a mAP of 0.67 has been achieved on real data. A novel metric for comparing works at the state-of-the-art (SOTA) in beach litter monitoring is also introduced, named "density-normalized F-score". The proposed methodology represents a benchmark for the definition of a standardize procedure for the indirect evaluation and monitoring of the coastal environmental status. Besides allowing the investigation of large areas with limited human effort, the proposed system enables the evaluation of the beach litter spatial distribution and magnitude, providing useful information for the assessment of tailored beach quality indices.
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11367/116576
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 23
  • ???jsp.display-item.citation.isi??? 19
social impact