Use of AI for SAR image classification on-board
Importance of the work
The space environment is changing and new actors in terms of data providers complement the space industry with low cost micro-satellites with the purpose to increase EO data. The use of AI is the solution to manage these increasing volumes of data in the near future.
ICEYE is a new company that is building a constellation of micro-satellites equipped with SAR. Its activities in terms of data quality and AI techniques are focused to reinforce the synergy with ESA, increasing the collaboration, providing data validated through the ESA guidelines, sharing data for common projects as well as ideas on different applications.
The use of AI is particularly relevant when applied to the remote sensing area. The first challenge is the exploitation of a very large amount of ICEYE X-band EO data to maximise the informative content while minimising the processing time. In this context, a dedicated dataset is to be prepared with the needed ground truth and this will be shared with the Φ-lab researchers.
The second challenge is the use of AI for on-board processing. In perspective, AI could reduce the volume of data which is required to be downlinked in the process and could allow the on-board implementation of dedicated detection algorithms.
The aforementioned challenges are dependent on the calibration and validation process that is currently on-going on ICEYE satellites X2, X4 and X5.
With the amount of EO data drastically increasing and the appearance of new companies with new micro-satellites ICEYE is helping to foster the implementation of processing techniques based on AI to enable the rapid processing of large amounts of data, thus reducing the amount of unnecessary data transmission to the ground.
At the same time, it is important that each EO data provider characterises the quality of the data provided to the users after the calibration and validation phase.
ICEYE activities go in this direction, performing a quality assessment on the X-band SAR data and developing on board and on ground processing techniques based on AI.
In the context of the data processing on ground, ICEYE will prepare a dedicated dataset with the available ground truth for land classification and maritime monitoring using semantic segmentation with NN and will be shared with the Ф-lab researchers.