Deep learning has achieved a remarkable performance breakthrough in several fields, most notably in speech recognition, natural language processing, and computer vision. In particular, convolutional neural network (CNN) architectures currently produce state-of-the-art performance on a variety of image analysis tasks such as object detection and recognition. Most of deep learning research has so far focused on dealing with 1D, 2D, or 3D Euclideanstructured data such as acoustic signals, images, or videos. Recently, there has been an increasing interest in geometric deep learning, attempting to generalize deep learning methods to non-Euclidean structured data such as graphs and manifolds, with a variety of applications from the domains of network analysis, computational social science, or computer graphics. In this paper, we propose a unified framework allowing to generalize CNN architectures to non-Euclidean domains (graphs and manifolds) and learn local, stationary, and compositional task-specific features. We show that various non-Euclidean CNN methods previously proposed in the literature can be considered as particular instances of our framework. We test the proposed method on standard tasks from the realms of image-, graphand 3D shape analysis and show that it consistently outperforms previous approaches.
In this article, we investigated the detection of forest vegetation changes during the period of 2017 to 2019 in the Low Tatras National Park (Slovakia) and the Sumava National Park (Czechia) using Sentinel-2 data. The evaluation was based on a time-series analysis using selected vegetation indices. The case studies represented five different areas according to the type of the forest vegetation degradation (one with bark beetle calamity, two areas with forest recovery mode after a bark beetle calamity, and two areas without significant disturbances). The values of the trajectories of the vegetation indices (normalized difference vegetation index (NDVI) and normalized difference moisture index (NDMI)) and the orthogonal indices (tasseled cap greenness (TCG) and tasseled cap wetness (TCW)) were analyzed and validated by in situ data and aerial photographs. The results confirm the abilities of the NDVI, the NDMI and the TCW to distinguish disturbed and undisturbed areas. The NDMI vegetation index was particularly useful for the detection of the disturbed forest and forest recovery after bark beetle outbreaks and provided relevant information regarding the health of the forest (the individual stages of the disturbances and recovery mode). On the contrary, the TCG index demonstrated only limited abilities. The TCG could distinguish healthy forest and the gray-attack disturbance phase; however, it was difficult to use this index for detecting different recovery phases and to distinguish recovery phases from healthy forest. The areas affected by the disturbances had lower values of NDVI and NDMI indices (NDVI quartile range Q2–Q3: 0.63–0.71; NDMI Q2–Q3: 0.10–0.19) and the TCW index had negative values (Q2–Q3: −0.06–−0.05)). The analysis was performed with a cloud-based tool—Sentinel Hub. Cloud-based technologies have brought a new dimension in the processing and analysis of satellite data and allowed satellite data to be brought to end-users in the forestry sector. The Copernicus program and its data from Sentinel missions have evoked new opportunities in the application of satellite data. The usage of Sentinel-2 data in the research of long-term forest vegetation changes has a high relevance and perspective due to the free availability, distribution, and well-designed spectral, temporal, and spatial resolution of the Sentinel-2 data for monitoring forest ecosystems.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.