Multi-Resolution Remote Sensing Dataset for the Detection of Anthropogenic Litter: A Multi-Platform and Multi-Sensor Approach

The dataset developed within the PlasticObs+ project aims to facilitate a multi-resolution approach for detecting and quantifying anthropogenic litter through areal images. Traditional detection methods often suffer from narrow, use-case-specific limitations, reducing their transferability. To address this, an image dataset was created featuring various spatial and spectral resolutions. The highest spatial resolution images (ground sampling distance = 0.2 cm) were used to generate a labeled dataset, which was georeferenced for mapping onto coarser-resolution images.