Time series-based machine learning algorithms for mapping dominant tree species in Flanders from Sentinel data

This presentation was given on the ForestSAT 2022 conference. 

It compares three machine learning and deep learning methods for the classification of dominant tree species in forests in Flanders from Sentinel-2 time series: Random Forest, 1D-CNN (Pelletier et al, 2019) and Time Series Forest (Deng et al, 2013). 

References

Deng, H., Runger , G., Tuv , E., & Vladimir, M. (2013). A time series forest for classification and feature extraction. Information Sciences , 239 , 142 153.

Pelletier, C., Webb, G., & Petitjean , F. (2019). Temporal Convolutional Neural Network for the Classification of Satellite Image Time Series. Remote Sensing , 11 (5), 523.