Session on Accelerating Decision Advantage (Sensors to shooters)

Chairperson and Speaker

Air Vice Marshal Sanjay Bhatnagar, VM, VSM (Retd)

Advisor | BEL | India

Defence Support System

Col Milan Patnaik

DGMI | India

Advances in Generative AI for SAR to EO Translation

EO (Electro-Optical) Satellite Imagery while extremely valuable, suffers from characteristic logistical issues such as loss of access under- (1) extreme weather conditions (cloud cover, disaster) and (2) low illumination (night-time). SAR (Synthetic Aperture Radar) imaging overcomes both these issues by the virtue of its underlying physics of operation. SAR offers all-season, all-weather imaging capabilities. However, the benefits of EO include high-definition scenes and high degree of human user friendly interpretability which is essential for Geoint applications. With the advances in generative AI techniques (example generative adversarial networks) researchers have explored the possibilities of artificially/synthetically generating representative EO images from SAR images. These techniques if implemented efficiently can act as a virtual force multiplier for requirements of space based situational-awareness. In this study we will outline and review state-of-the-art-techniques and their performance for SAR to EO translation/generation which can have multiple applications for defense and civil use-cases.