• Long Short-Term Memory Neural Network for Sentinel-2: This architecture can either be trained using samples selected by the user or be trained directly on the TimeSen2Crop dataset to perform the crop type classification. The service include both the training phase and the inference.

• Long Short-Term Memory Neural Network for NDVI prediction. This architecture can be trained using an user-generated time series of NDVI values. The network trains on the user's input, and predicts the NDVI values on new acquisitions.
• Pre-Trained Long Short-Term Memory Neural Network: A pre-trained version of the LSTM on the TimeSen2Crop dataset is available in .h5 format. This network can be used to classify the specified tile harmonized using the monthly composite approach.

- Training and inference on user custom datasets.

- Possibility of fine-tuning the pre-trained network, to exploit the trained weights.

- Code packaged in a Docker image and the services can be used as a Docker container.