DySENSE is a joint multimodal GenAI framework that generates static urban satellite imagery and spatially aligned dynamic monthly building energy maps, with conditioning on urban morphology, dynamic energy use, and climate cues. Experiments across four cities show that the framework generates high-fidelity synthetic energy data using less than 25% labeled data, improves downstream predictive performance by 8% mIoU, and delivers an 8.7% accuracy gain over existing methods while reducing CVRMSE by 9%–18% (ASHRAE).
Source:Hugging Face 社区论文精选 · Read Original
Leave a Reply