Structured light point cloud from the downloadable dataset (set6). This data set (freely available) is aimed at multiple view stereo (MVS) evaluation, and is made using our robotic lab set up outlined here. An industrial robot arm was mounted with a structured light scanner.

Jan 29, 2023 · DTU是一个 MVS数据集，常被被用来做MVS、 NeRF方法 的训练集。 DTU [1]是一个大型数据集，包含受控实验室环境中的128个场景，其模型使用 结构光扫描仪 捕获。 在7种不同的照明条件下，在相同的49或64个相机位置扫描每个场景，生成分辨率为1200×1600像素的RGB图像。 由于数据集涵盖了各种对象和材料，因此非常适合在现实条件下训练和测试深度学习MVS方法。 因为参考模型是点云，地面真实深度图应根据网格模型进行渲染，网格模型是使用表面重 …

Each scene consists of 49 or 64 accurate camera positions and reference structured light scans, all acquired by a 6-axis industrial robot. DTU MVS 2014 is a multi-view stereo dataset, which is an order of magnitude larger in number of scenes and with a significant increase in diversity.

At present we have two available data sets: One aimed at evaluating point features, made in 2010 with a journal publication appearing in 2012. One aimed at evaluating multiple view stereo, made in 2013 and published in 2014.

To start the reconstruction for the skull, run: or for a general scan: You will find the output meshes in the directory ./out/{SCAN-ID}_{SCAN-NAME}/meshes. The DTU dataset in the NDS format is derived from the dataset in IDR format (found here), which includes masks for …

May 6, 2024 · 本文介绍了DTU数据集的结构，包括用于训练和评估的128个扫描，详细的分组（train,val,test），相机参数、深度图真值以及不同光照条件下的rectified图像。 此外，还提及了点云GT用于定量测试的评测指标。

A python implementation of DTU MVS 2014 evaluation. It only takes 1min for each mesh evaluation. And the gap between the two implementations is negligible.

The json representation of the dataset with its distributions based on DCAT. Junshu Tang, Tengfei Wang, Bo Zhang, Ting Zhang, Ran Yi, Lizhuang Ma, Dong Chen (2024). Dataset: DTU 3D Scan Dataset. https://doi.org/10.57702/qrken82s.

Dec 23, 2023 · We evaluate our proposed method both qualitatively and quantitatively on the DTU dataset and show high surface reconstruction quality with relatively accurate camera poses compared with other baselines.

Version update 2024-02-16: Changed test data to also be from european scans. The FDI 16 tooth mesh dataset including original meshes and exact data used in the original VF-Net paper. All scans are fully anonymized and donated by 3Shape.