IonQ researchers tested a quantum generative model, QCBM, using real satellite radar imagery from Capella Space. The experiments utilized IonQ Forte and Forte Enterprise processors to analyze high-resolution synthetic aperture radar (SAR) data, including images of the Marine Corps Air Station Miramar and the Piton de la Fournaise volcano.
The quantum model demonstrated superior performance on complex datasets with sparse pixel statistics, achieving an F1-score of 0.32 compared to 0.16 and 0.24 for classical methods. On an ideal simulator, the QCBM score reached 0.41. However, when data was pre-processed to improve statistical distribution, the performance gap between quantum and classical approaches narrowed.
The study highlights that the quantum approach is particularly effective for high-resolution imagery where traditional methods struggle due to insufficient observations. Researchers noted that the model currently requires separate training for each pair of images, though preliminary tests showed potential for cross-site application without retraining.
IonQ intends to continue exploring quantum models for infrastructure monitoring, environmental analysis, and defense applications. This follows the company's recent announcement regarding the deployment of a 256-qubit Superion 256 system at the NVIDIA Accelerated Quantum Research Center to develop hybrid quantum-classical computing systems.
