PSIque: Next Sequence Prediction of Satellite Images using a Convolutional Sequence-to-Sequence Network
This work addresses disaster management by improving weather prediction from satellite data, but it appears incremental as it builds on existing sequence-to-sequence and skip connection techniques.
The paper tackles the problem of predicting future satellite images for weather forecasting by proposing a convolutional sequence-to-sequence autoencoder with symmetric skip connections, achieving the closest predictions to ground truth images in experiments.
Predicting unseen weather phenomena is an important issue for disaster management. In this paper, we suggest a model for a convolutional sequence-to-sequence autoencoder for predicting undiscovered weather situations from previous satellite images. We also propose a symmetric skip connection between encoder and decoder modules to produce more comprehensive image predictions. To examine our model performance, we conducted experiments for each suggested model to predict future satellite images from historical satellite images. A specific combination of skip connection and sequence-to-sequence autoencoder was able to generate closest prediction from the ground truth image.