Modern synthetic aperture radar systems are capable of capturing wide swaths at very high resolutions, resulting in an immense amount of raw data to process. Even after image reconstruction, transmitting the complex imagery from the platform to a secondary processing site for object extraction can present significant challenges, especially when the platform is subject to emission control measures. This presentation will address methods for performing automatic object recognition on low SWaP-C systems and the associated challenges with respect to accuracy and robustness.