Bi-Directional LSTM Model For Classification Of Vegetation From Satellite Time Series

To further improve the classification accuracy of remote sensing time series data, in this paper, we propose a Bi-directional Long-Term and Short-Term Memory (denoted as BI-LSTM) based model for vegetation mapping and monitoring. The proposed model is applied to multi-temporal publicly available Sentine1-2A dataset with vegetation as the main theme. Experimental results have shown that the proposed approach has good performance in comparison with the state-of-the-art methods in term of accuracy, precision and recall. Moreover, it can efficiently use both past and future input features using the BI-LSTM component.