Browse Teach Curate

Global-Detail Dual-Branch Network with RWKV for Remote Sensing Image Super-Resolution

A super-resolution architecture for satellite and aerial imagery that pairs a linear-complexity RWKV branch for global context with a convolutional branch for local detail, fused through a dedicated reconstruction module. A student learns how to combine global and local feature modeling under tight compute budgets and how a wavelet-domain loss enforces cross-resolution frequency fidelity.

Data Text

JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 1 GDSR: Global-Detail Integration through

GDSR (Global-Detail dual-branch network) is a deep-learning model for super-resolving remote sensing images. It runs two parallel branches - a Receptance Weighted Key Value (RWKV) branch that capture…