UAV ACTUAL: 3D CNN

Mostrando entradas con la etiqueta 3D CNN. Mostrar todas las entradas

domingo, 8 de noviembre de 2020

3D Fire Front Reconstruction in UAV-Based Forest-Fire Monitoring System

This work presents a new method of 3D reconstruction of the forest-fire front based on uncertain observations captured by remote sensing from UAVs within the forest-fire monitoring system.

The use of multiple cameras simultaneously to capture the scene and recognize its geometry including depth is proposed. Multi-directional observation allows perceiving and representing a volumetric nature of the fire front as well as the dynamics of the fire process.

The novelty of the proposed approach lies in the use of soft rough set to represent forest fire model within the discretized hierarchical model of the terrain and the use of 3D CNN (3D Convolutional Neural Network) to classify voxels within the reconstructed scene.

The developed method provides sufficient performance and good visual representation to fulfill the requirements of fire response decision makers.

domingo, 11 de octubre de 2020

Classification of Grassland Desertification in China Based on Vis-NIR UAV Hyperspectral Remote Sensing

In this study, a vis-NIR (visual Near Infra Red) hyperspectral remote sensing system for UAVs (Unmanned Aerial Vehicles) was used to analyze the type and presence of vegetation and soil of typical desertified grassland in Inner Mongolia using a DBN (Deep Belief Network), 2D CNN (2D Convolutional Neural Network) and 3D CNN (3D Convolutional Neural Network).

The results show that these typical deep learning models can effectively classify hyperspectral data on desertified grassland features. The highest classification accuracy was achieved by 3D CNN, with an overall accuracy of 86.36%. This study enriches the spatial scale of remote sensing research on grassland desertification, and provides a basis for further high-precision statistics and inversion of remote sensing of grassland desertification.