Dataset Information

PASMVS: A perfectly accurate, synthetic, path-traced dataset featuring specular material properties for multi-view stereopsis training and reconstruction applications.

ABSTRACT: A Perfectly Accurate, Synthetic dataset for Multi-View Stereopsis (PASMVS) is presented, consisting of 400 scenes and 18,000 model renderings together with ground truth depth maps, camera intrinsic and extrinsic parameters, and binary segmentation masks. Every scene is rendered from 45 different camera views in a circular pattern, using Blender's path-tracing rendering engine. Every scene is composed from a unique combination of two camera focal lengths, four 3D models of varying geometrical complexity, five high definition, high dynamic range (HDR) environmental textures to replicate photorealistic lighting conditions and ten materials. The material properties are primarily specular, with a selection of more diffuse materials for reference. The combination of highly specular and diffuse material properties increases the reconstruction ambiguity and complexity for MVS reconstruction algorithms and pipelines, and more recently, state-of-the-art architectures based on neural network implementations. PASMVS serves as an addition to the wide spectrum of available image datasets employed in computer vision research, improving the precision required for novel research applications.

SUBMITTER: Broekman A

PROVIDER: S-EPMC7474405 | biostudies-literature | 2020 Oct

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

PASMVS: A perfectly accurate, synthetic, path-traced dataset featuring specular material properties for multi-view stereopsis training and reconstruction applications.

Broekman André A Gräbe Petrus Johannes PJ

Data in brief 20200824

A Perfectly Accurate, Synthetic dataset for Multi-View Stereopsis (PASMVS) is presented, consisting of 400 scenes and 18,000 model renderings together with ground truth depth maps, camera intrinsic and extrinsic parameters, and binary segmentation masks. Every scene is rendered from 45 different camera views in a circular pattern, using Blender's path-tracing rendering engine. Every scene is composed from a unique combination of two camera focal lengths, four 3D models of varying geometrical com ...[more]

PMID: 32923541

Similar Datasets

Project description:This dataset contains the material volumes, masses, and intensities for a total of 45 residential building cohorts in Finland from the 1940s to the 2010s. The specific building types included are one dwelling houses and blocks of flats. The data were drawn from representative case buildings and their derivatives. The data are primarily based on construction drawings, complemented by other documents such as bills of materials. The source material was mainly obtained from the archives of the building inspection authority of the city of Vantaa, Finland. Material volumes were derived from the construction drawings either directly from annotations or, when needed, by further measurements made based on the same material. For minor lacks of information in the original documents, documents of similar buildings and literature were consulted. A total of 26 buildings were inventoried directly. For each included combination of building type, construction decade, and bearing material these were the ones with the most common façade material. In addition, 19 buildings with the second most common façade material were formed based on these to represent the 45 cohorts. Material masses, and by extension intensities, were calculated based on the recorded volumes and typical densities of construction materials used in Finland. The material volumes, masses, and intensities per material and in total are presented as three spreadsheet tables, along with a description sheet, on three corresponding hierarchical levels of aggregation: per representative building, per vertical building level (foundations, basement, first storey, etc.), and per building part (floor, exterior walls, interior walls, etc.). Furthermore, they are distinguished between the building structure and complementary building components (windows and doors). The data can be used in academic, policy related, and practical investigations of the building stock, such as in evaluating the material consumption consequences of different spatial planning strategies on various levels or estimating the materials embedded in the built environment and their potential for capitalisation in the circular economy.

Dataset Information

PASMVS: A perfectly accurate, synthetic, path-traced dataset featuring specular material properties for multi-view stereopsis training and reconstruction applications.

Publications

PASMVS: A perfectly accurate, synthetic, path-traced dataset featuring specular material properties for multi-view stereopsis training and reconstruction applications.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets