Off the shelf deep learning pipeline for remote sensing applications
R. Tripathi, A. Chan-Hon-Tong and A. Boulch
Published in Big Data from Space (BIDS), ESA Workshop, 2017
Designing specific index for a some remote sensing applications require a large research effort not scalable to the multitude of applications.
Inversely, using off the shelf deep learning pipeline could be good enough for some applications.
We describe off the shelf deep learning application on the 2017 data fusion contest (IEEE-IGARSS) for local climate zone estimation. While being completely non expert to local climate zone estimation, and while having only few meta parameters, these pipelines reach honorable scores on this dataset compared to hard to tune winner pipeline of the challenge.