SatReg: Regression-based Neural Architecture Search for Lightweight Satellite Image Segmentation
For researchers deploying remote-sensing segmentation on edge platforms, SatReg offers a practical method to tune architecture parameters without exhaustive search, though the results are incremental and domain-specific.
SatReg proposes a regression-based neural architecture search method for lightweight satellite image segmentation on edge devices, reducing the search space to two width variables and using surrogate models to predict mIoU, latency, and power. The approach enables fast selection of near-optimal architectures, achieving efficient adaptation of hybrid CNN-Mamba models for space-edge systems.
As Earth-observation workloads move toward onboard and edge processing, remote-sensing segmentation models must operate under tight latency and energy constraints. We present SatReg, a regression-based hardware-aware tuning framework for lightweight remote-sensing segmentation on edge platforms. Using CM-UNet as the teacher architecture, we reduce the search space to two dominant width-related variables, profile a small set of student models on an NVIDIA Jetson Orin Nano, and fit low-order surrogate models for mIoU, latency, and power. Knowledge distillation is used to efficiently train the sampled students. The learned surrogates enable fast selection of near-optimal architecture settings for deployment targets without exhaustive search. Results show that the selected variables affect task accuracy and hardware cost differently, making reduced-space regression a practical strategy for adapting hybrid CNN-Mamba segmentation models to future space-edge systems.