Join us at the JACIE 2024 Workshop

GXP’s own Todd Jobe will be presenting Enhancing Satellite Image Resolution with Deep Learning, Wednesday 13 March at 1pm.

In this session, attendees will discover a new system developed by BAE Systems GXP that applies deep learning to enhance satellite image resolutions. A super-resolution model leveraging Deep Convolution Neural Networks (DCNNs) is trained using individual image bands, a large crop size or tile size of 512 x 512 pixels, and a de-noise algorithm.

This super-resolution system for multispectral and hyperspectral geospatial satellite imagery can deduce higher resolution images from lower resolution images while maintaining the original color of the lower resolution pixels. Comparison studies of automatic object detection accuracy between the original images and the super-resolution images indicate that with super-resolution images generated by this system, the object detection accuracy (recall and precision) has improved by 20% with multispectral satellite images.