depthschatting
Depthschatting, also known as depth estimation, is the process of determining the distance from a viewpoint to surfaces in a scene, producing a depth map that encodes per-pixel depth or a 3D representation such as a point cloud. Depthschatting can be performed from images, video, or dedicated sensors.
Geometric methods include stereo matching, where depth is triangulated from pairs of images with a known baseline;
Sensor fusion combines data from cameras with LiDAR or other ToF/structured-light sensors to improve accuracy, robustness,
Representations include depth maps (per-pixel depth), disparity maps (inverse depth from stereo), point clouds, and 3D
Applications span robotics, autonomous driving, augmented reality and virtual reality, 3D reconstruction, geographic surveying, and navigation.
Evaluation and datasets use metrics such as RMSE, mean absolute error, and absolute relative error, along with