DISCO-2 image processing and machine learning pipeline

The DISCO-2 satellite will have accelerated machine learning capability based on the inclusion of a Coral TPU ML accelerator module. This will allow images taken by the satellite to be analaysed on satellite using a variety of ML models, with only select images sent back to Earth. This approach allows for more flexibility in image aquisition and saves downlink bandwidth which is very constrained in the case of DISCO-2.

This project will address the entire image pipeline including pre-processing of images on the satellite after aquisition for insertion into the ML pipeline, the selection of images based on the results of ML analysis, the transfer of images to the groundstation and the data storage and presentation of the images.

The project extends previous work done for DISCO-1 and includes work on linux on embeded devices, cubesat space protocol/libcsp, image databases and implemented primarily in python and C languages. There is an opportunity to work with other universities in the program.