Bi-temporal satellite method cuts building damage data sent after disasters
Embedded Bi-Temporal Building Damage Assessment for On-Board Data Reduction
Computer Vision and Pattern RecognitionArtificial Intelligence
Summary
After natural disasters, quickly knowing which buildings are damaged helps emergency teams respond faster. The authors developed a way for satellites to send only important damage details, not full images, by comparing new pictures with compressed older ones on board. Their method keeps detection accuracy even with compressed data and fixes alignment problems between images. They tested it on specialized computers in satellites and found it works efficiently.
What this means in practice
- •For disaster response teams: Rapidly receive only key building damage data from satellites to prioritize emergency actions after disasters.
- •For satellite system engineers: Implement onboard data reduction and alignment techniques to reduce communication loads and speed processing of disaster images.
Authors
Thomas Goudemant, Benjamin Francesconi, Marjorie Bellizzi, Adrien Dorise
Abstract
Rapid assessment of building damage after natural disasters is essential to support emergency response. Earth Observation satellites can acquire relevant imagery shortly after an event, but exploitation is limited by uplink and downlink capacity and by ground-processing latency. We address this with a bi-temporal building damage assessment pipeline built on a siamese detector derived from YOLOX, designed to compress information at both ends of the ground/space link. On the ground, pre-disaster reference images are encoded into a compact latent space -- compressed by up to a factor of 64 -- and uplinked to the satellite. On board, this reference is compared with a fresh post-disaster acquisition so that the downlink carries only actionable object-level products, bounding boxes and damage classes, instead of full scenes. This cuts the data exchanged in both directions, while on xBD the strongly compressed reference still preserves most of the detection performance. Because on-board acquisitions suffer from residual pre/post co-registration errors, we introduce a latent-space shift estimation and correction module that regresses the global offset from the coarse feature level and realigns the post-disaster features before fusion. It substantially improves robustness to de-registration -- especially under large shifts, where fusion-only variants collapse -- while also raising nominal accuracy and remaining compatible with the strongest compression. We finally port the pipeline to two embedded targets, a Xilinx Versal VCK190 and an NVIDIA Jetson AGX Orin, and report hardware performance (latency, throughput, power efficiency). The core detector and its compression port cleanly to both, but the operators needed for long-range robustness survive only on the Jetson GPU, whereas the Versal DPU does not.