Large Scale Multi-view Stereopsis Evaluation

By Rasmus Ramsbøl Jensen, Anders Bjorholm Dahl, George Vogiatzis, Engil Tola, Henrik Aanæs. 2014 IEEE Conference on Computer Vision and Pattern Recognition, 2014

1,030 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/736baa2d935ebffca0683bdc729dff36eba15d68

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

Citations in the corpus

DNGaussian: Optimizing Sparse-View 3D Gaussian Radiance Fields with Global-Local Depth Normalization (2024, methodology): “The DTU dataset [17] consists of 124 object-centric scenes captured by a set of fixed cameras.”
PGSR: Planar-Based Gaussian Splatting for Efficient and High-Fidelity Surface Reconstruction (2024, neutral): “To model the overall brightness variations at different times, we assign two exposure coefﬁcients, a and b , to each image.”
ReconFusion: 3D Reconstruction with Diffusion Priors (2023, positive): “6, our test set performance is consistently better than Zip-NeRF, indicating that our diffusion prior can serve as an effective drop-in regularizer across a range of capture settings.”
Towards Probabilistic Volumetric Reconstruction Using Ray Potentials (2015, background): “More recently, researchers have shifted their focus from reconstructing isolated single objects to more challenging general scenes that involve significant occlusions, textureless or reflective surfaces, transient objects, varying illumination conditions, and camera mis-registration errors [7, 16, 1”
Multiview Neural Surface Reconstruction by Disentangling Geometry and Appearance (2020, methodology): “We apply our multiview surface reconstruction model to real 2D images from the DTU MVS repository [15].”
VGGT: Visual Geometry Grounded Transformer (2025, positive): “Following MASt3R [43], we further evaluate our multi-view depth estimation results on the DTU [35] dataset.”
Computer Vision for Autonomous Vehicles: Problems, Datasets and State-of-the-Art (2017, methodology): “In contrast, the TUD MVS dataset12 by Jensen et al. (2014) provides 124 different scenes that were also recorded”
MVSNeRF: Fast Generalizable Radiance Field Reconstruction from Multi-View Stereo (2021, methodology): “We train our framework on the DTU [20] dataset to learn a generalizable network.”
MVSplat: Efficient 3D Gaussian Splatting from Sparse Multi-View Images (2024, methodology): “Furthermore, to further assess the cross-dataset generalization ability, we also directly evaluate all models on the multi-view DTU [17] dataset, which contains object-centric scenes with camera poses.”
Volume Rendering of Neural Implicit Surfaces (2021, methodology): “DTU The DTU [12] dataset contains multi-view image (49 or 64) of different objects with fixed camera and lighting parameters.”
Multi-View Mesh Reconstruction with Neural Deferred Shading (2022, neutral): “We test our method on the DTU multiview dataset [22] with the object selection and masks from previous work [41, 64].”
RegNeRF: Regularizing Neural Radiance Fields for View Synthesis from Sparse Inputs (2021, methodology): “Datasets We report results on the real world multi-view datasets DTU [21] and LLFF [35].”
pixelNeRF: Neural Radiance Fields from One or Few Images (2020, methodology): “recognition, as well as domain transfer to real car photos and 3) real scenes from the DTU MVS dataset [14].”
Deep Marching Cubes: Learning Explicit Surface Representations (2018, result): “We follow common practice [22] and specify all measures as distances, thus lower accuracy / completeness values indicate better results.”
NeRF++: Analyzing and Improving Neural Radiance Fields (2020, background): “The higher complexity required for incorrect shapes is more difﬁcult to represent with a limited capacity MLP. (Jensen et al., 2014), replacing NeRF’s MLP with a vanilla MLP signiﬁcantly reduces generalization to novel views.”
ScanNet++: A High-Fidelity Dataset of 3D Indoor Scenes (2023, methodology): “Meanwhile, datasets that were originally proposed for multi-view stereo such as DTU [15], BlendedMVS [47], and Tanks and Temples [20] are now also used for novel view synthesis.”