Panoptic Segmentation of Environmental UAV Images : Litter Beach

EasyChair Preprint 14236

Abstract

Convolutional neural networks (CNN) have been used efficiently in several fields, including environmental research. In fact, CNN can help with monitoring of marine litter, which has become a worldwide problem. UAVs have a higher resolution and more adaptable in local area than satellite imagery, which makes it easier to find and count trash. Since the sand is heterogeneous, a simple CNN model encounters plenty of inferences caused by reflections of sand color, human footsteps, shadows, algae present, dunes, holes, and tire tracks. For this type of image, segmentation methods based on CNN can be more appropriate. In this paper, we use an instance-based segmentation method and a panoptic segmentation method that show good accuracy with just a few datasets. The model is more robust and less sensitive to disturbances, especially in panoptic segmentation.

Keyphrases: Convolutional Neural Network CNN, Litter Beach, environment, panoptic segmentation, unmanned aerial vehicle UAV

@booklet{EasyChair:14236,
  author    = {Ousmane Youme and Jean Marie Dembele and Eugene C. Ezin and Christophe Cambier},
  title     = {Panoptic Segmentation of Environmental UAV Images : Litter Beach},
  howpublished = {EasyChair Preprint 14236},
  year      = {EasyChair, 2024}}