ScanNet++: A High-Fidelity Dataset of 3D Indoor Scenes

By Chandan Yeshwanth, Yueh-Cheng Liu, Matthias Nießner, Angela Dai. IEEE International Conference on Computer Vision, 2023

795 citations recorded by the source corpus.

We present ScanNet++, a large-scale dataset that couples together capture of high-quality and commodity-level geometry and color of indoor scenes. Each scene is captured with a high-end laser scanner at sub-millimeter resolution, along with registered 33-megapixel images from a DSLR camera, and RGB-D streams from an iPhone. Scene reconstructions are further annotated with an open vocabulary of semantics, with label-ambiguous scenarios explicitly annotated for comprehensive semantic understanding. ScanNet++ enables a new real-world benchmark for novel view synthesis, both from high-quality RGB capture, and importantly also from commodity-level images, in addition to a new benchmark for 3D semantic scene understanding that comprehensively encapsulates diverse and ambiguous semantic labeling scenarios. Currently, ScanNet++ contains 460 scenes, 280,000 captured DSLR images, and over 3.7M iPhone RGBD frames.

Subfields: Computational Mechanics, Computer Graphics and Computer-Aided Design, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/fc96c4a7d4708bfd6138bfd16482229975404499

How researchers see this paper

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

In Computational Mechanics

In Computer Graphics and Computer-Aided Design

In Computer Vision and Pattern Recognition

Citations in the corpus

DL3DV-10K: A Large-Scale Scene Dataset for Deep Learning-based 3D Vision (2023, neutral): “RealEstate10k [42], ARKitScenes [5], ScanNet [11], and the high-resolution ScanNet++ [36] improve this with a broad range of detailed indoor scenes.”
Thinking in Space: How Multimodal Large Language Models See, Remember, and Recall Spaces (2024, positive): “These videos are sourced from the validation sets of the public in-door 3D scene reconstruction datasets ScanNet [19], Scan-Net++ [97], and ARKitScenes [5] and represent diverse en-vironments—including residential spaces, professional settings ( e.g ., offices, labs), and industrial spaces ( e.g .,…”
Depth Anything 3: Recovering the Visual Space from Any Views (2025, positive): “…Synthetic Omniobject [102] 5885 Synthetic OmniWorld [128] 1039 Synthetic PointOdyssey [127] 44 Synthetic ReplicaVMAP [83] 17 Synthetic ScanNet++ [117] 230 LiDAR ScenenetRGBD [55] 16866 Synthetic TartanAir [98] 355 Synthetic Trellis [104] 557408 Synthetic vKitti2 [9] 50 Synthetic WildRGBD [103]…”
Fast3R: Towards 3D Reconstruction of 1000+ Images in One Forward Pass (2025, positive): “While image pairs are an important use case, often one is interested in re-constructing from more than two views, as when scanning of objects [39] or scenes [4, 6, 20, 55, 67], e.g . for asset generation or mapping.”
Grounding Image Matching in 3D with MASt3R (2024, methodology): “We train our network with a mixture of 14 datasets: Habitat [74], ARKitScenes [20], Blended MVS [112], MegaDepth [48], Static Scenes 3D [57], ScanNet++ [113], CO3D-v2 [67], Waymo [83], Map-free [5], WildRgb [2], VirtualKitti [12], Unreal4K [91], TartanAir [103] and an internal dataset.”
DUSt3R: Geometric 3D Vision Made Easy (2023, positive): “Our model is trained in a fully-supervised manner using a simple regression loss, leveraging large public datasets for which ground-truth annotations are either synthetically generated [69, 104], reconstructed from SfM softwares [56, 162] or captured using dedicated sensors [25, 94, 122, 166].”