1

We propose a method that combines a Fully Convolutional Network and a Graph Neural Network to efficiently extract road networks from aerial images in a single pass.

This paper presents a method that captures resolution-free object contours as complex periodic functions.

Using 1D convolutional LSTM to efficiently process satellite images of arbitrary length.

We present and evaluate different strategies for the binarization of graph neural networks. We show that through careful design of the models, and control of the training process, binary graph neural networks can be trained at only a moderate cost in accuracy.

Designing lightweight encoder-decoder networks using depth-wise separable convolutions for onboard satellite image analysis.