Interpretable vision-language reasoning for diagnosing estate-scale residential renewal urgency

Assessing the urban renewal urgency enables planners to prioritize interventions and allocate resources effectively. Current visual-based evaluation methods rely primarily on pattern recognition and fail to provide explicit diagnostic reasoning about observed conditions. Vision-language models (VLMs) offer a promising approach by incorporating intrinsic semantic knowledge and enabling interpretable visual reasoning. Therefore, this study combines a VLM with chain-of-thought (CoT) reasoning to assess renewal urgency from observable physical environmental conditions at the housing estate level. Using Beijing as a case study, we develop multidimensional assessment criteria and fine-tune Qwen3-VL-8B through low-rank adaptation (LoRA) with CoT supervision. The adapted model improves performance across most assessment dimensions. For overall renewal urgency, it achieves an F1-score of 0.760 and a quadratic weighted kappa (QWK) of 0.785. These represent relative improvements of 13.1% and 6.5% over the baseline without CoT. The trained model classifies 2947 housing estates (56.6%) as having moderate overall renewal urgency. Renewal urgency decreases from the city center toward outer ring road zones, reflecting the concentration of aging infrastructure in older inner-city areas. Furthermore, the model provides human-readable reasoning outputs that articulate dimension-specific visual evidence, improving transparency and practical utility. The framework provides a scalable and evidence-based approach to interpretable image-based assessment of urban renewal urgency.