利用遥感手段对人类活动强度开展量化评估，对于统筹协调南岭北麓古村落群生态保护与城乡建设行为具有重要意义。基于2012年、2016年、2020年以及2024年四期高分一号卫星数据，运用U-Net分类器提取土地利用类型信息，并基于多年土地利用转移矩阵和人类活动强度综合指数对研究区169个古村落进行人类活动强度评估。研究结果表明：(1)国土调查数据和年度变更调查数据为主的遥感解译样本和U-Net模型相融合，可实现研究区人类活动用地类型的高精度监测提取；(2)南岭北麓古村落群自2010年来，以建设用地为主的人类活动现象总体呈现加速的态势，古村落区域人类活动强度综合指数均值从1.73增长至3.55，平均年增长为17.10%，并存在显著的空间分异；(3)土地利用转移矩阵信息体现古村落群内从农田、自然植被向建设用地的主转移方向，但存在10.5%的古村落区域出现逆转移现象。本研究聚焦南岭北麓古村落群，创新性构建了融合高分辨率遥感与深度学习的人类活动强度动态评估框架，揭示了文化遗产地-人-地关系的双向演变规律，为应对快速城镇化背景下的传统村落保护提供了科学性与可操作性的方案。

Quantitative assessment of human activity intensity by remote sensing means is of great significance for coordinating the ecological protection and urban and rural construction of ancient villages at the northern foot of Nanling Mountains. Based on the GF-1 satellite data in2012, 2016, 2020 and 2024, we used U-Net classifier to extract the land-use type information, and used multi-year land-use transfer matrix and the comprehensive index of human activity intensity to assess the human activity intensity of 169 ancient villages in the study area. The research results indicate that(1)the fusion of remote sensing interpretation samples mainly based on territorial survey data and annual change survey data with U-Net model can achieve high-precision monitoring and extraction of human activity land-use types in the study area.(2)Since 2010, the ancient villages at the northern foot of Nanling Mountains have witnessed an acceleration of human activities, mainly on construction land. The average comprehensive index of human activity intensity in the ancient villages has increased from 1.73 to 3.55, with an average annual growth of 17.10%, and there is significant spatial differentiation.(3) The land-use transfer matrix information reflects the main transfer direction from farmland and natural vegetation to construction land within the ancient village cluster, but there is a reverse transfer phenomenon in 10.5% of the ancient village areas. Focusing on the ancient villages at the northern foot of Nanling Mountains, we innovatively constructed a dynamic assessment framework of human activity intensity integrating high-resolution remote sensing and deep learning, revealed the both way evolution law of human and land relationship in cultural heritage sites, and provided a scientific and operable scheme for the protection of traditional villages in the context of rapid urbanization.