Researchers have developed a novel deep-learning framework for estimating building ages using readily available multi-source geospatial data. This approach overcomes limitations of existing methods by integrating remote sensing data, derived products, and building footprint information. The framework employs a large pretrained backbone with fine-tuning adapters and specialized fusion modules for both pixel-wise and instance-wise predictions, incorporating a temporal ordinal constraint to enforce age sequencing. A new benchmark dataset for Hong Kong has also been created to facilitate comprehensive evaluation. AI
IMPACT This framework could improve urban planning and infrastructure safety by enabling more accurate and scalable building age estimation.
RANK_REASON This is a research paper detailing a new deep-learning framework for building age estimation. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
