CJRS’ Special Issue, Volume 47 Issue 2

Deep Learning for Environmental
Applications of Remote Sensing Data

Inspired by human brains’ great potential for object and pattern recognition, deep learning (DL) has drawn attention within the remote sensing community over the past few years. Supervised convolutional neural networks (CNNs), recurrent neural networks (RNNs), unsupervised auto-encoders (AE), deep belief networks (DBNs), and generative adversarial networks (GANs) are the state-of-the-art DL methods applied for remote sensing imagery. DL methods’ popularity is attributed to their deep multilayer structure, allowing the extraction of robust, invariant, and high-level features of data and their end-to-end training scheme. In other words, these methods have the capability to learn a series of abstract hierarchical features from raw input data and to provide a final, task-specific output, thus removing heuristic feature engineering design. Continue reading …

Guest Editors
Masoud Mahdianpari, Saeid Homayouni & Samuel Foucher