Abstract:Despite significant progress made in the past few years, challenges remain for depth estimation using a single monocular image. First, it is nontrivial to train a metric-depth prediction model that can generalize well to diverse scenes mainly due to limited training data. Thus, researchers have built large-scale relative depth datasets that are much easier to collect. However, existing relative depth estimation models often fail to recover accurate 3D scene shapes due to the unknown depth shift caused by training with the relative depth data. We tackle this problem here and attempt to estimate accurate scene shapes by training on large-scale relative depth data, and estimating the depth shift. To do so, we propose a two-stage framework that first predicts depth up to an unknown scale and shift from a single monocular image, and then exploits 3D point cloud data to predict the depth shift and the camera's focal length that allow us to recover 3D scene shapes. As the two modules are trained separately, we do not need strictly paired training data. In addition, we propose an image-level normalized regression loss and a normal-based geometry loss to improve training with relative depth annotation. We test our depth model on nine unseen datasets and achieve state-of-the-art performance on zero-shot evaluation. Code is available at: https://github.com/aim-uofa/depth/.

3D Information Acquisition from Rectangles and Circles in Single Images

A Simple 3D Reconstruction Method Based on Single Camera

Multi-view 3D Reconstruction from Uncalibrated Radially-Symmetric Cameras

A Flexible Method for 3D Reconstruction of Urban Building

Precise 3d reconstruction from a single image

Sketch-based 3-D modeling for piecewise planar objects in single images

Unwrapping and Stereo Rectification for Omnidirectional Images

Pose and Position Estimations of Regular Shapes in Monocular Images

Centra Catadioptric Camera Calibration with Single Image.

3D Ultrasound Reconstruction Based on Rotational Scanning Coping with Calibration Uncertainty

Metric Planar Rectification from Perspective View Via Circles.

Distance Measurement from Single Image Based on Circles

3D reconstruction from bifocus imaging

Efficient Extraction of Metric Measurements for Planar Scene under 2D Homography with the Help of Planar Circles

Stereo Image Rectification Algorithm for Multi-View 3D Display

3d Stereoscopic Image Generation From Monocular View In Structured Environments

Understanding of Indoor Scenes Based on Projection of Spatial Rectangles.

Towards Accurate Reconstruction of 3D Scene Shape From A Single Monocular Image

3D Reconstruction from a Single Still Image Based on Monocular Vision of an Uncalibrated Camera

Efficient 3D Reconstruction Using Monocular Vision

A Novel 3d Reconstruction Method from A Single Uncalibrated Camera