Large-Scale Data for Multiple-View Stereopsis

By Henrik Aanæs, Rasmus Ramsbøl Jensen, George Vogiatzis, Engin Tola, Anders Bjorholm Dahl. 2016

758 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition, Geology

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

Citations in the corpus

RayNet: Learning Volumetric 3D Reconstruction with Ray Potentials (2018, methodology): “Our second dataset is the widely used DTU multi-view stereo benchmark [1], which comprises 124 indoor scenes of various objects captured from 49 camera views under seven different lighting conditions.”
SurfaceNet: An End-to-End 3D Neural Network for Multiview Stereopsis (2017, methodology): “Figure 1: Reconstruction of the model 13 of the DTU dataset [1] in comparison to [4, 8, 27, 9].”
Recurrent MVSNet for High-Resolution Multi-View Stereo Depth Inference (2019, methodology): “Table 1: Quantitative results on the DTU evaluation scans [1].”
PatchmatchNet: Learned Multi-View Patchmatch Stereo (2020, methodology): “We use the model trained on DTU [1] without any ﬁne-tuning.”
Fast3R: Towards 3D Reconstruction of 1000+ Images in One Forward Pass (2025, neutral): “We hypothesize that the additional context provided by more views during training allows the model to learn higher-order correspondences between multiple frames, ultimately increasing model performance and increasing potential for scaling.”
NeuralRecon: Real-Time Coherent 3D Reconstruction from Monocular Video (2021, background): “Learning-based approaches in MVS have recently dominated several benchmarks [2, 20] in terms of accuracy, but are only limited to processing mid-resolution images due to the GPU memory constraint.”
DUSt3R: Geometric 3D Vision Made Easy (2023, neutral): “Following [110], we evaluate it on the DTU [1], ETH3D [109], Tanks and Temples [50], and ScanNet [23] datasets.”
BlendedMVS: A Large-Scale Dataset for Generalized Multi-View Stereo Networks (2019, methodology): “It has been reported on the small object DTU dataset [2] that, the best overall quality has been largely improved by recent learning-based approaches [34, 35, 4, 13].”
MonoSDF: Exploring Monocular Geometric Cues for Neural Implicit Surface Reconstruction (2022, methodology): “Thus we consider: a) Real-world indoor scans: Replica [67] and ScanNet [13]; b) Real-world large-scale indoor scenes: Tanks and Temples [34] advanced scenes; c) Object-level scenes: DTU [1] in the sparse 3-view setting from [46,84].”
Point-Based Multi-View Stereo Network (2019, methodology): “The DTU dataset [1] is a large-scale MVS dataset, which consists of 124 different scenes scanned in 7 different lighting conditions at 49 or 64 positions.”
Neural volumes (2019, background): “12 shows the results of the experiment on four objects we captured as well as one scene from a publicly available MVS dataset [Aanæs et al. 2016].”
Objaverse-XL: A Universe of 10M+ 3D Objects (2023, positive): “Similar to pretraining in 2D vision and language, we observe that pretraining on Objaverse-XL with PixelNeRF improves performance when fine-tuning to other datasets such as DTU [61] and ShapeNet [13] (Table 3).”
Deep Stereo Using Adaptive Thin Volume Representation With Uncertainty Awareness (2019, methodology): “We compare the accuracy and the completeness of the final reconstructions using the distance metric in [1].”
Image Matching from Handcrafted to Deep Features: A Survey (2020, background): “DTU (Aanæs et al. 2016): The dataset is originally designated for multiple-view stereo evaluation, which involves a number of different scenes with a wide range of objects.”
Make-It-3D: High-Fidelity 3D Creation from A Single Image with Diffusion Prior (2023, methodology): “We evaluate the proposed method on public DTU dataset [1] and our benchmark, and extensive experiments show a clear improvement over previous works.”
Cost Volume Pyramid Based Depth Inference for Multi-View Stereo (2019, methodology): “DTU Dataset [1] is a large-scale MVS dataset with 124 scenes scanned from 49 or 64 views under 7 different lighting conditions.”