Structure-from-Motion Revisited

By Johannes L. Schönberger, Jan-Michael Frahm. 2016

7,734 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition, Geology, 3D Vision & Pose

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Aerospace Engineering

In Computer Vision and Pattern Recognition

In Geology

In 3D Vision & Pose

Citations in the corpus

Multiview Compressive Coding for 3D Reconstruction (2023, methodology): “CO3D extracts point clouds from the videos via COLMAP [54, 55], which are inevitably noisy and are used to train our model.”
Pixel2Mesh: Generating 3D Mesh Models from Single RGB Images (2018, background): “The major research directions include structure from motion (SfM) [27] for large-scale high-quality reconstruction and simultaneous localization and mapping (SLAM) [4] for navigation.”
OctNetFusion: Learning Depth Fusion from Data (2017, background): “While image based techniques [40, 15, 12, 49, 39] provide compelling results for sufficiently textured surfaces, the introduction of affordable depth sensors, in particular the Microsoft Kinect sensor, allows scanning a wide variety of objects and scenes.”
Visual Imitation Made Easy (2020, methodology): “ons’ taken by the human demonstrator is crucial. To address this, we recover 6DoF poses of the tool using Structure-from-Motion (SfM) reconstruction. Speciﬁcally, we use the publicly available COLMAP [44, 45] software for SfM. Once we have the end-effector pose p tfor every image I t, we extract the”
Cognitive Mapping and Planning for Visual Navigation (2017, methodology): “lable. When using RGB images as input, we triangulated SIFT feature points in the RGB images (registered using the ground truth egomotion) to obtain the observed 3D points (we used the COLMAP library [67]). This occupancy map was used to compute a grid-graph (unoccupied cells are assumed free). For ”
Learning 3D Object Categories by Looking Around Them (2017, methodology): “We extract this information using off-theshelf methods: the structure-from-motion (SFM) algorithm COLMAP for RGB sequences [33, 34], and an open-source implementation [31] of KinectFusion (KF) [24] for RGBD sequences.”
Kornia: an Open Source Differentiable Computer Vision Library for PyTorch (2019, background): “Local feature detectors and descriptors Local features are the workhorses of 3d reconstruction [49, 45], visual localization [48] and image retrieval [52].”
Atlas: End-to-End 3D Scene Reconstruction from Posed Images (2020, methodology): “We compare our method to 4 state-of-the-art baselines: COLMAP [42, 43], MVDepthNet [51], GPMVS [26], and DPSNet [28].”
DeepVoxels: Learning Persistent 3D Feature Embeddings (2018, background): “Model-Based Rendering Classic reconstruction approaches such as structure-from-motion exploit multi-view geometry [15, 50] to build a dense 3D point cloud of the imaged scene [46, 47, 49, 1, 12].”
SOSNet: Second Order Similarity Regularization for Local Descriptor Learning (2019, methodology): “The process of describing local patches is a fundamental component in many computer vision tasks such as 3D reconstruction [31, 33], large scale image localization [30] and image retrieval [29].”
GS-SLAM: Dense Visual SLAM with 3D Gaussian Splatting (2023, methodology): “In contrast, we derive the analytical derivative equations for pose estimation in the Gaussian representation and implement efficient CUDA optimization.”
GS-LRM: Large Reconstruction Model for 3D Gaussian Splatting (2024, methodology): “To address this task, various traditional methods have been proposed, including structure-from-motion (SfM) [1, 48, 50, 56] for sparse reconstruction and calibration, and multi-view stereo (MVS) [23,47,52] for dense reconstruction.”
Generative Novel View Synthesis with 3D-Aware Diffusion Models (2023, methodology): “For (3), we provide COLMAP [64, 65] reconstructions of generated video sequences, a standard evaluation for 3D consistency in 3D GANs [66, 7, 6].”
Compact 3D Gaussian Representation for Radiance Field (2023, methodology): “3DGS constructs initial 3D Gaussians derived from the sparse data points obtained by Structure-from-Motion (SfM), such as COLMAP [37].”
NeRF++: Analyzing and Improving Neural Radiance Fields (2020, methodology): “This dataset consists of hand-held 360 ◦ captures of four large-scale scenes; the camera poses are estimated by COLMAP SfM (Sch¨onberger & Frahm, 2016).”
A Cordial Sync: Going Beyond Marginal Policies for Multi-Agent Embodied Tasks (2020, methodology): “Methods entailing structure-from-motion and SLAM [91,80,25,13,72,81] were used to build maps.”
Nerfstudio: A Modular Framework for Neural Radiance Field Development (2023, methodology): “Taichi [Hu et al. 2019], and Reyes [Cook et al. 1987] for graphics, Phototourism [Snavely et al. 2006] and COLMAP [Schönberger and Frahm 2016; Schönberger et al. 2016a,b] for photogrammetry and visualization, and AverageExplorer [Zhu et al. 2014] for data collection.”