Building Change Detection

Building change detection and monitoring are critical activities in urban development, infrastructure maintenance, public safety, and environmental sustainability. In urban settings, where the city expands and evolves, it is important to have methods for identifying any changes in buildings, whether it is newly built constructions, building alterations, or demolition.

Currently, conventional building change detection is achieved using manual surveys, inspections, or visual analysis of imagery. All these methods are inefficient as they are laborious and require a lot of time, not to mention scalability. Modern technological developments, such as computer vision, deep learning, and high-resolution aerial images, allow for identifying building changes efficiently.

The proposed solution is a machine learning model that detects building changes using high-resolution aerial or drone images and creates a continuous change magnitude raster. Instead of creating just a binary raster, where a pixel can either be “changed” or “unchanged”, the model generates change probability/magnitude per pixel.

This continuous raster output will make it easier to detect any slight changes that might not be possible to notice during manual examination, such as additions to buildings, demolishing parts, constructions, roof adjustments, and other changes within the built environment. This information can be further used and integrated into the GIS platform to enable visualization, analysis, monitoring, and decision-making.

The application of building change detection using deep learning has numerous uses in different areas, such as monitoring urban growth, zoning and planning, updating property records, monitoring construction, assessing infrastructure, detecting potentially unauthorized construction, and analyzing other aspects. In addition, this method can also help authorities understand how natural events like floods, earthquakes, storms, and fires have impacted buildings and the urban environment around them.

Due to the automated processing of large amounts of aerial images, the use of deep learning and drones to detect changes in urban areas becomes more efficient and consistent at the same time.