TU Wien Research Data
Not a member yet
512 research outputs found
Sort by
RT-Percept Lumberyard Bistro
No full text<p>Pre-rendered dataset used in <a href="https://jaliborc.github.io/rt-percept/">Training and Predicting Visual Error for Real-Time Applications</a> for the Lumberyard Bistro scenes. Generated using the RT-Percept renderer and the RT-Percept scenes.</p>
SilviLaser 2021 Benchmark Dataset - Terrestrial Challenge
No full text<p>This benchmark dataset was acquired during the SilviLaser conference 2021 in Vienna. The benchmark aims to demonstrate the different terrestrial system's capabilities for capturing 3D scenes in various forest conditions. A number of universities, institutes, and companies participated and contributed their outputs to this dataset, compiled by terrestrial laser scanning (TLS), mobile laser scanning (MLS), as well as terrestrial photogrammetric systems (TPS). Along with the terrestrial data, one airborne laser scanning (ALS) data was provided as a reference.</p><p>Eight forest plots were installed in the terrestrial challenge. Each plot was formed with a 25-meter radius circular area and different tree species (i.e. spruce, pine, beech, white fir), forest structures (i.e. one layer, multi-layer, natural regeneration, deadwood), and age classes (~50 – 120 years). The 3D point clouds acquired by each participant cover the eight plots. In addition to point clouds, traditional in-situ data (tree position, tree species, DBH) were recorded by the organization team.</p><p>All point clouds provided by participants were processed in the following steps: co-registration with geo-referenced data, setting a uniform coordinate reference system (CRS), and removing data located out of the plot. This work was performed by OPALS, a laser scanning data processing software developed by the Photogrammetry Group of the TU Wien Department of Geodesy and Geoinformation. Please note that some point clouds are not archived due to problems encountered during pre-processing. The final products consist of one metadata, 3D point clouds, ALS data for reference, and corresponding digital terrain models (DTM) derived from the ALS data using OPALS software. Point clouds are in laz 1.4 format, and DTMs are raster models in GeoTIFF format. Furthermore, all geo-data use CRS of WGS84 / UTM zone 33N (EPSG:32633). More information (e.g. instrument, point density, and extra attributes) can be found in the file "SL21BM_TER_metadata.csv".</p><p>This dataset is available to the community for a wide variety of scientific studies. These unique data sets will also form the basis for an international benchmark for parameter retrieval from different 3D recording methods.</p><h2>Acknowledgements</h2><p>This dataset was contributed by the universities/institutes/companies (alphabetical order):</p><ul><li>Czech University of Life Sciences Prague</li><li>Forest Design</li><li>Green Valley International</li><li>RIEGL</li><li>Silva Tarouca Research Institute</li><li>Swiss Federal Institute for Forest, Snow and Landscape Research</li><li>Umweltdata GmbH</li><li>University of Natural Resources and Life Sciences</li><li>Wageningen University & Research</li></ul><h2>Notes</h2><ol><li>In terms of in-situ data, please contact <a href="mailto:[email protected]">Markus Hollaus</a> for details.</li><li>To perform a bulk download, please use <a href="https://owncloud.tuwien.ac.at/index.php/s/J8tMtk1BQH0hBhL/download">this file</a> to get the URL list.</li></ol>
European Sentinel-1 Forest Type and Tree Cover Density Maps
No full text<p>This dataset was generated by the <a href="https://mrs.geo.tuwien.ac.at/">TU Wien Department of Geodesy and Geoinformation</a>.</p><p>European Sentinel-1 forest type and tree cover density maps represent first continental-scale forest layers based on Sentinel-1 C-Band Synthetic Aperture Radar (SAR) backscatter data. For the year 2017 they cover the majority of European continent with 10 m and 100 m sampling for forest type and tree cover density, respectively. The maps were derived using the method described in <a href="https://www.tandfonline.com/doi/full/10.1080/01431161.2018.1479788">https://www.tandfonline.com/doi/full/10.1080/01431161.2018.1479788</a>.</p><p>The forest type map shows the dominant forest type class (coniferous, broadleaf). Tree cover density map shows the percentage of forest canopy cover within the 100 m pixel.</p><p>Please be referred to our peer-reviewed article at <a href="https://doi.org/10.3390/rs13030337">https://doi.org/10.3390/rs13030337</a> for details and accuracy assessment accross Europe.</p><h1>Dataset Record</h1><p>The forest type and tree cover density maps are sampled at 10 m and 100 m pixel spacing respectively, georeferenced to the Equi7Grid and divided into square tiles of 100km extent ("T1"-tiles). With this setup, the forest maps consist of 728 tiles over the European continent, with data volumes of 3.12 GB and 378.3 MB.</p><p>The tiles' file-format is a LZW-compressed GeoTIFF holding 16-bit integer values, with tagged metadata on encoding and georeference. Compatibility with common geographic information systems as QGIS or ArcGIS, and geodata libraries as GDAL is given.</p><p>In this repository, we provide each forest map as tiles, whereas two zipped dataset-collections are available for download below.</p><h1>Code Availability</h1><p>For the usage of the <strong>Equi7Grid</strong> we provide data and tools via the python package available on GitHub at <a href="https://github.com/TUW-GEO/Equi7Grid">https://github.com/TUW-GEO/Equi7Grid</a>. More details on the grid reference can be found in <a href="https://www.sciencedirect.com/science/article/pii/S0098300414001629">https://www.sciencedirect.com/science/article/pii/S0098300414001629</a>.</p><h1>Acknowledgements</h1><p>The computational results presented have been achieved using the Vienna Scientific Cluster (VSC).</p>
Daily snow cover grid maps over Austria in the period 2000-2020
No full text<p>The dataset provides daily snow cover grid maps for the territory of Austria in the period 2000-2020. The dataset consists of two products based on observations from Terra (in the period 2000-2020) and Aqua (period 2002-2020) satellites, both using Moderate Resolution Imaging Spectroradiometer (MODIS) data. The spatial resolution of all maps is 500m. </p><p>The snow cover maps for Austria are processed from MODIS, Version 006, <a href="https://doi.org/10.5067/MODIS/MOD10A1.006">MOD10A1</a> (Terra, Hall and Riggs, 2016) and <a href="https://doi.org/10.5067/MODIS/MYD10A1.006">MYD10A1</a> (Aqua, Hall and Riggs, 2016) products. These products provide daily global grid maps of a Normalized Difference Snow Index (NDSI), which represents a difference in reflectance observations in the visible and infrared MODIS bands (band 4 and band 6). Snow cover classification requires then attributing the NDSI values to snow or land classes. The dataset uses for the snow cover classification a method developed and introduced in <a href="https://doi.org/10.1016/j.jhydrol.2020.125548">Tong et al. (2020)</a>. In comparison, previous approaches used a spatially and temporally fixed NDSI (i.e. if the NDSI value for a pixel was larger than 0.4, then the pixel was classified as snow-covered). The study of Tong et al. (2020) found that for snow cover mapping in Austria, the accuracy can be (3-10% larger) if the NDSI threshold varies spatially and seasonally. The overall accuracy of snow cover mapping in Austria is larger than 97%. Table 1, which is attached as a file in <i>Table1_NDSI_thresholds.csv</i>, shows the final criteria used for snow cover mapping as found for Austria in Tong et al., 2020.</p><p>The snow cover maps provide for each 500x500m grid information, whether it is classified as snow-covered or not, or it is not possible to classify it due to cloud cover. The limitation of the optical sensing of snow cover is that the clouds block the observation at the ground. The availability of daily observations from two satellites allows using several cloud reduction techniques if needed. More information about the cloud reduction approaches and their effectiveness in Austria can be found in <a href="https://doi.org/10.1029/2007WR006204">Parajka et al. (2008)</a> or <a href="https://doi.org/10.1016/j.jhydrol.2014.08.064">Krajci et al. (2014)</a>. </p><p>The format of the grid maps is GeoTiff. The images are projected from WGS84 using the Lambert Conformal Conic projection (LCC_WGS_1984). Each image covers an area within the boundaries of (10 000, 275 000) meters and (770 000, 715 000) meters (bottom-left, top-right).</p><p>The possible grid values and their corresponding classes are summarized below in <i>Table2_Grid_value_classification.csv</i>.</p><p>File Naming Convention includes name of the satellite product (Terra or Aqua), date in the format year (YYYY) and acquisition day of year (DDD). For example: </p><ul><li><i>Terra_2017003_snowcovermap_Austria.tif</i> refers to a grid map from Terra satellite from 3rd January 2017.</li><li><i>Aqua_2007365_snowcovermap_Austria.tif</i> refers to a grid map from Aqua satellite from 31th December 2007.</li></ul>
The CLEF-IP 2012 Test Collection
No full text<h3>CLEF-IP: Cross-Language Evaluation Forum - Intellectual Property</h3><p>The CLEF-IP track ran from 2009 to 2013 and aimed to investigate IR techniques for patent retrieval.The track utilizes a collection of more than 1.3M patent documents (~2.6 million files) derived from EPO (European Patent Office) sources and EuroPCT Applications (more than 400K documents) published by WIPO (World Intelectual Property Organization). The collection contains documents in English, French and German with at least 150,000 documents in each language, all published before 2001.</p><p>There were three tasks in 2012: The first one was to find patent documents that are candidates to constitute prior art for a given claim taken from a patent document. The second task, flowchart recognition, asked participants to extract the information in these images and return it in a predefined textual format. The third task, chemical structure regonition, participants had to identify the location of the chemical structures depicted on images of patent pages and, for each of them, return the corresponding structure in a MOL file (a chemical structure file format).</p><p>Files</p><ol><li><strong>Document Collection</strong><br>The first one is a set of XML files representing a total of over 1.3 million patent documents.<br>NOTE: the document collection is the same as the one published for CLEF-IP 2011, excluding images.</li><li><strong>Topics and Answers</strong><br>Both the training and the test topic sets contain also the relevance assessments for the topics.</li></ol><p> </p>
The CLEF-IP 2009 Test Collection
No full text<h3>CLEF-IP: Cross-Language Evaluation Forum - Intellectual Property</h3><p>The CLEF-IP track was launched in 2009 to investigate IR techniques for patent retrieval and it is part of the CLEF 2009 evaluation campaign.The track utilizes a collection of more than 1M patent documents derived from EPO (European Patent Office) sources. The collection contains documents in English, French and German with at least 100,000 documents in each language. The task is to find patent documents that constitute prior art. The topics are complete patent documents that participants can process to extract queries. In addition to the Main task, CLEF-IP 2009 provided three language tasks (English, German, French) where topics were in one of these three languages.</p><p>Relevance judgements were produced by two methods: automatically, using patent citations from seed patents; and manual for a small number of queries for which search results will be reviewed by Intellectual Property Experts.</p><h3>Files</h3><ol><li><strong>Document Collection</strong><br>The CLEF-IP 2009 collection of documents consists of XML files. There are 1,9 million XML files, corresponding to approximately 1 million individual patents filed between 1985 and 2000. A dtd file for the XML format is provided as well.</li><li><strong>Topics and Answers (Qrels)</strong><br>Both the training and the test topic sets contain also the relevance assessments for the topics. For each task of the CLEF-IP 09 track, we provide 4 sets of different sizes of topic test sets: XLarge, Large, Medium, Small.</li><li><strong>Guidelines</strong><br>Contains detailed explanation on how to work with the four tasks from the corpus.</li></ol><h3><br> </h3>
The CLEF-IP 2011 Test Collection
No full text<h3>CLEF-IP: Cross-Language Evaluation Forum - Intellectual Property</h3><p>The CLEF-IP track ran from 2009 to 2013 and aimed to investigate IR techniques for patent retrieval.The track utilizes a collection of more than 1.3M patent documents (~2.6 million files) derived from EPO (European Patent Office) sources and EuroPCT Applications (more than 400K documents) published by WIPO (World Intelectual Property Organization). The collection contains documents in English, French and German with at least 150,000 documents in each language, all published before 2001.</p><p>There were four tasks in 2011: The first one was to find patent documents that are candidates to constitute prior art for a given document. The second task was to classify a given document according to the International Patent Classification system (IPC). The third task was to retrieve patent document that are candidates for prior art, where, in addition to the first task, images contained in the patent documents must be proceesd as well. The fourth task was to classifiy patent images into 9 classes in use at patent office. Relevance judgements are produced using the patent citations and meta-data.</p><h3>Files</h3><ol><li><strong>Document Collection</strong><br>The corpus consists of two parts. The first one is a set of XML files representing a total of over 1.3 million patent documents - this collection is to be used for the first task. The second part is a subset of the first one to which images containing patent drawings are added - this collection is to be used for the third task. </li><li><strong>Topics and Answers</strong><br>Both the training and the test topic sets contain also the relevance assessments for the topics.</li><li><strong>Guidelines</strong><br>Detailed explanation on how to work with the tasks from the corpus.</li></ol>
The FAIR Principles
No full text<p>According to the <a href="https://www.go-fair.org/fair-principles/">FAIR Principles</a>, research objects should be findable, accessible, interoperable and re-usable. These principles form the basis of a trusted environment where researchers, innovators, companies, and citizens can publish, find, and re-use each other's data and tools for research, innovation, and educational purposes. The FAIR principles refer to any digital object evolving from the research process, that is, quantitative as well as qualitative data, metadata or algorithms, tools, software and services.</p><p>This presentation summarises what the FAIR Principles mean and how they can be applied. </p><p> </p>
The MAREC/IREC data set
No full text<h3>MAREC/IREC: The MAtrixware REsearch Collection / The Information retrieval facility Research Collection</h3><p>MAREC/IREC is a static collection of over 19 million patent applications and granted patents in a unified file format normalized from EP, WO, US, and JP sources, spanning a range from 1976 to June 2008. MAREC/IREC is intended as raw material for research and evaluation in areas such as information retrieval, natural language processing or machine translation, which require large amounts of complex documents. It allows experiments with real data on a realistic scale.<br>The collection contains documents in several languages, the majority being English, German and French, and about half of the documents include full text.</p><p>In MAREC/IREC, the documents from different countries and sources are normalized to a common XML format with a uniform patent numbering scheme and citation format. The standardized fields include dates, countries, languages, references, person names, and companies as well as rich subject classifications. It is a comparable corpus, where many documents are available in similar versions in other languages.</p><p>The 19,386,697 XML files measure a total of 621 GB. </p><h3>IREC - Information retrieval facility Research Collection</h3><p>The MAREC original collection was missing parts of the European Granted Patents claim section (EP-B documents). An EPB_Bugfix folder existed to provide those files corrected. The IREC simply merges the original EPB folder with the EPB_Bugfix in order to provide a uniform representation. The CLEF-IP collections have never been affected by this issue, as they were specially curated.</p><h3>License Information</h3><p>MAREC by IRF is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License. Permissions beyond the scope of this license may be available at mailto:[email protected].</p><h3> </h3>
Virtual Annotated Cooking Environment Dataset
No full text<p>This dataset was recorded in the Virtual Annotated Cooking Environment (VACE), a new open-source virtual reality dataset (https://sites.google.com/view/vacedataset) and simulator (https://github.com/michaelkoller/vacesimulator) for object interaction tasks in a rich kitchen environment. We use the Unity-based VR simulator to create thoroughly annotated video sequences of a virtual human avatar performing food preparation activities. Based on the MPII Cooking 2 dataset, it enables the recreation of recipes for meals such as sandwiches, pizzas, fruit salads and smaller activity sequences such as cutting vegetables. For complex recipes, multiple samples are present, following different orderings of valid partially ordered plans. The dataset includes an RGB and depth camera view, bounding boxes, object masks segmentation, human joint poses and object poses, as well as ground truth interaction data in the form of temporally labeled semantic predicates (holding, on, in, colliding, moving, cutting). In our effort to make the simulator accessible as an open-source tool, researchers are able to expand the setting and annotation to create additional data samples.</p><p>The research leading to these results has received funding from the Austrian Science Fund (FWF) under grant agreement No. I3969-N30 InDex and the project Doctorate College TrustRobots by TU Wien. Thanks go out to Simon Schreiberhuber for sharing his Unity expertise and to the colleagues at the TU Wien Center for Research Data Management for data hosting and support.</p><p> </p>