A lightweight CNN called ConvBEERS restores satellite images with 6.9 dB PSNR gain, boosts downstream object detection by up to 5.1% mAP, and deploys on Xilinx Versal FPGA with 41x lower latency than physics-based pipelines.
Garc´es-Socarr´as, Raudel Cuiman, Flor Ortiz, Juan A
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Rethinking Satellite Image Restoration for Onboard AI: A Lightweight Learning-Based Approach
A lightweight CNN called ConvBEERS restores satellite images with 6.9 dB PSNR gain, boosts downstream object detection by up to 5.1% mAP, and deploys on Xilinx Versal FPGA with 41x lower latency than physics-based pipelines.