Collected molecules will appear here. Add from search or explore.
A model-hardware co-design framework for Synthetic Aperture Radar (SAR) Automatic Target Recognition (ATR) that optimizes CNNs for adversarial robustness and inference efficiency on resource-constrained hardware like satellites.
citations
0
co_authors
4
This project occupies a highly specialized niche (SAR ATR for satellites). While the codebase has zero stars and is likely a research artifact, the domain expertise required for SAR-specific data (complex-valued signals, speckle noise) and hardware-level optimization provides a natural moat against general-purpose AI labs. Frontier labs are unlikely to compete in orbital edge computing for radar imaging.
TECH STACK
INTEGRATION
reference_implementation
READINESS