I give some variations of this courses at ENSG, the French school of geomatic, as well as at conferences tutorials (JURSE2019, ISPRS Congress 2021), online (EURSODR-EduSERV 2019-2020), and as training for researchers (NIBIO Oslo in 2019, LaSTIG in 2021).

The courses requires a familiarity with Python.

Deep Learning for Remote Sensing Course

1) Introduction to Machine Learning for Remote Sensing:
Content: degrees of supervision, classification vs regression vs clustering, common metrics, classic algorithms (SVM, RF, K-means)
Exercise: tree species classification from morphological and geographical features

2) Introduction to Deep Learning for Remote Sensing:
Content: principles of deep learning, neural architecture, neural networks in practice
Exercise: crop type classification from multispectral satellite images

3) Deep Learning for Image:
Content: training a neural network, extended layer bestiary, advanced convolutional architectures
Exercise: pixel-precise land use segmentation from multispectral aerial images

4) Deep Learning for 3D Point Clouds:
Content: review of the 3D litterature, scaling segmentation algorithms, latest advances
Exercise: land cover segmentation from aerial LiDAR 3D scans

5) Deep Learning for Time Series (coming in 2022):
Content: lit review: Pre-deep & RNNs & Transformers, specificty of remote sensing applications, presentation of recent high-performing architectures for remote sensing
Exercise: crop type classification from multispectral satellite images time series