Multimodal Modular Chain of Thoughts in Energy Performance Certificate Assessment

Researchers developed a Multimodal Modular Chain of Thoughts (MMCoT) framework using Vision-Language models to automate Energy Performance Certificate assessments from visual data. Testing on 81 UK residential properties showed significant improvements over traditional prompting methods, offering a cost-effective solution for energy efficiency evaluation in data-scarce regions.

• →MMCoT architecture decomposes EPC estimation into intermediate reasoning stages for better accuracy.
• →The framework achieved statistically significant improvements over instruction-only prompting methods.
• →Testing on 81 UK residential properties demonstrated the system's ability to capture ordinal structure of EPC ratings.
• →Most prediction errors occurred between adjacent rating classes, indicating reasonable performance boundaries.
• →The approach offers a promising low-cost solution for energy performance assessment in regions lacking scalable EPC systems.