3D Reconstruction
The process of building three-dimensional models of objects or scenes from 2D images or sensor data.
Overview
3D reconstruction is the task of creating three-dimensional geometric models from two-dimensional observations such as photographs, video, or depth sensor data. Techniques range from classical multi-view stereo and structure-from-motion to modern deep learning approaches like NeRF and 3D Gaussian Splatting.
Key Details
Deep learning has dramatically improved 3D reconstruction quality, enabling single-image 3D prediction and real-time scene reconstruction. Applications span autonomous driving, robotics, augmented/virtual reality, medical imaging, architecture, and cultural heritage digitization.