Haiying Wang and Mingzhong Wu, researchers from China West Normal University, published a study on May 24, 2026, in Scientific Reports detailing a deep learning approach to urban change detection in hilly regions. The research focuses on Nanchong, a city in Sichuan Province, China, to address challenges in urban sustainable development within complex landscapes characterized by fragmented terrain and scattered buildings.

The study involved three primary tasks. First, researchers developed two region-specific datasets: the Building and Road Semantic Segmentation Dataset (BR_Data_NC) and the Change Detection Dataset (CD_Data_NC). These were specifically designed to support deep learning models in interpreting medium-resolution remote sensing imagery typical of hilly terrains.

Second, the team applied deep learning models to perform semantic segmentation (a process where each image pixel is classified into categories like building or road) of existing urban features and to detect changes over time using the constructed datasets. The experiments showed that these deep learning methods effectively manage spectrally mixed features and landscape fragmentation, which often complicate analysis in hilly urban areas.

Third, the researchers utilized the detected urban changes to analyze expansion patterns, revealing evolutionary trends in the city's spatial morphology. The findings suggest that deep learning provides a reliable tool for dynamic monitoring, which can assist in optimizing land-use efficiency and informing spatial planning for sustainable development in western China's hilly cities.

The study was funded by the Doctoral Start-up Fund Project of China West Normal University, under project number 23KE007. The authors declared no competing interests regarding the research. The paper, received on January 31, 2026, and accepted on May 14, 2026, is currently available as an unedited version before its final publication.

Haiying Wang and Mingzhong Wu, researchers from China West Normal University, published a study on May 24, 2026, in Scientific Reports detailing a deep learning approach to urban change detection in hilly regions. The research focuses on Nanchong, a city in Sichuan Province, China, to address challenges in urban sustainable development within complex landscapes characterized by fragmented terrain and scattered buildings.

The study involved three primary tasks. First, researchers developed two region-specific datasets: the Building and Road Semantic Segmentation Dataset (BR_Data_NC) and the Change Detection Dataset (CD_Data_NC). These were specifically designed to support deep learning models in interpreting medium-resolution remote sensing imagery typical of hilly terrains.

Second, the team applied deep learning models to perform semantic segmentation (a process where each image pixel is classified into categories like building or road) of existing urban features and to detect changes over time using the constructed datasets. The experiments showed that these deep learning methods effectively manage spectrally mixed features and landscape fragmentation, which often complicate analysis in hilly urban areas.

Third, the researchers utilized the detected urban changes to analyze expansion patterns, revealing evolutionary trends in the city's spatial morphology. The findings suggest that deep learning provides a reliable tool for dynamic monitoring, which can assist in optimizing land-use efficiency and informing spatial planning for sustainable development in western China's hilly cities.

The study was funded by the Doctoral Start-up Fund Project of China West Normal University, under project number 23KE007. The authors declared no competing interests regarding the research. The paper, received on January 31, 2026, and accepted on May 14, 2026, is currently available as an unedited version before its final publication.