Contribution to a conference proceedings

FZJ-2021-01285

http://join2-wiki.gsi.de/foswiki/pub/Main/Artwork/join2_logo100x88.png Super-Resolution of Large Volumes of Sentinel-2 Images with High Performance Distributed Deep Learning

Zhang, R.FZJ* ; Cavallaro, G. (Corresponding author)FZJ* ; Jitsev, J.FZJ*

2020

This record in other databases:  

Please use a persistent id in citations: http://hdl.handle.net/2128/27350  doi:10.1109/IGARSS39084.2020.9323734

Abstract: This work proposes a novel distributed deep learning modelfor Remote Sensing (RS) images super-resolution. High PerformanceComputing (HPC) systems with GPUs are used toaccelerate the learning of the unknown low to high resolutionmapping from large volumes of Sentinel-2 data. The proposeddeep learning model is based on self-attention mechanismand residual learning. The results demonstrate that stateof-the-art performance can be achieved by keeping the size ofthe model relatively small. Synchronous data parallelism isapplied to scale up the training process without severe performanceloss. Distributed training is thus shown to speed uplearning substantially while keeping performance intact.

---

Contributing Institute(s):

1. Jülich Supercomputing Center (JSC)

Research Program(s):

1. 512 - Data-Intensive Science and Federated Computing (POF3-512) (POF3-512)
2. HBP - The Human Brain Project (604102) (604102)

Appears in the scientific report 2021

---

Database coverage:
; ; BIOSIS Previews ; Current Contents - Life Sciences ; Ebsco Academic Search ; IF < 5 ; JCR ; NCBI Molecular Biology Database ; SCOPUS ; Science Citation Index Expanded ; Thomson Reuters Master Journal List ; Web of Science Core Collection ; Zoological Record

---