Computer vision (CV) is a field of artificial intelligence that enables computers to extract, analyze, and understand information from digital images and videos. In earthquake engineering, it is increasingly used for automated crack detection, post-earthquake damage assessment, and structural component recognition.
Computer vision (CV) is a field of artificial intelligence that develops methods to acquire, process, analyze, and understand digital images and video. The goal is to enable machines to interpret visual information in a way that mimics human perception β recognizing objects, detecting patterns, measuring dimensions, and inferring three-dimensional structure from two-dimensional images. The field combines techniques from image processing, pattern recognition, machine learning, and geometry, and has been dramatically transformed over the past decade by deep learning β particularly convolutional neural networks (CNNs), which can learn hierarchical features directly from image data.
Computer vision is increasingly applied in earthquake engineering where visual inspection is slow, expensive, or dangerous. The main applications fall into several categories. Crack detection and quantification in concrete and masonry structures uses image processing and deep learning to identify cracks, measure their width and length, and track their evolution over time β replacing or supplementing manual inspection. Post-earthquake damage classification uses satellite, aerial, or drone imagery to classify buildings into damage states (no damage, slight, moderate, extensive, collapse) across large areas, enabling rapid situation assessment. Structural component recognition uses computer vision to identify columns, beams, walls, and other structural elements from images or point clouds, supporting as-built documentation and digital twin creation. Structural displacement monitoring uses photogrammetry or video-based tracking to measure the displacement of structures under earthquake or wind loading without contact β an alternative to traditional sensors. UAV-based inspection uses drones equipped with cameras and computer vision to inspect bridges, dams, and tall buildings where access is difficult. Photogrammetry and 3D reconstruction create three-dimensional models of structures from multiple images, supporting condition assessment and finite element model generation.
Computer vision in earthquake engineering is rapidly evolving and increasingly integrated with structural health monitoring and post-disaster response. The main technical challenges include: (1) variability in lighting, angle, and image quality that affects detection accuracy; (2) the need for large, labeled datasets β a persistent challenge for damage detection, where damaged-structure imagery is less common than intact-structure imagery; (3) transferability of models across different structures, materials, and regional construction practices; (4) the gap between pixel-level damage detection and engineering-level damage assessment (e.g., relating crack width to structural capacity); (5) real-time deployment for rapid post-earthquake assessment, which requires fast algorithms and reliable field hardware. Modern research increasingly combines computer vision with other sensing modalities β thermal imaging, LiDAR, and structural monitoring β to provide more complete information. Deep learning approaches dominate current research, with architectures including CNNs for classification and detection, U-Net and Mask R-CNN for semantic and instance segmentation, and vision transformers for more complex tasks. In Iran, computer vision research for earthquake engineering is growing rapidly, with applications in crack detection from imagery, damage classification from drone and satellite data, and displacement monitoring. Iran's growing UAV capabilities and the availability of high-resolution satellite imagery make CV particularly promising for large-scale post-earthquake assessment.