Loading...
Loading...
3D models, rendered datasets, physics simulation, digital twins, synthetic data generation, game engine data
1,020 datasets
TaskGrasp-Pro extends the original TaskGrasp dataset with fine-grained part decompositions and part-level physical property annotations for 190 household objects. The dataset, created by WCL-Robotics, defines three task types per object instance, resulting in a total of 2,850 tasks. It includes point clouds, multi-view RGB-D images, and task-related data.
Real-world point cloud recordings of a folding and unfolding cloth, used to evaluate state-estimation models from UniClothDiff. Each rollout contains two synchronized point cloud streams derived from an Intel RealSense depth camera and a monocular RGB image. The dataset was authored by Cloth-splatters and last updated on Hugging Face in May 2026.
tw-instruct-500k-2511 is a 2025 November version of a synthetic dialogue dataset for training Taiwanese Mandarin conversational models. It combines reference-based and reference-free generation methods to produce instruction-response pairs aligned with Taiwanese context. The dataset was created by lianghsun and updated on HuggingFace in May 2026.
Aggregating monthly zooplankton biomass samples from nine reference stations around the Australian coastline. The sampling program analyzes community composition, biomass, and size spectrum, with data planned for interoperability through IMOS infrastructure.
A 1500-year chronology of shoreline sediment accumulation is reconstructed from beach ridge deposits in Queensland, Australia. The Australian Ocean Data Network provides this study, which analyzes ridge morphology, sediment texture, and geochemistry. Preliminary results indicate changes in sediment supply rates, source areas, and minor relative sea-level falls.
Keppel Bay, Queensland, Australia, is the location for this study of a beach-ridge plain. The dataset likely contains measurements of ridge morphology, sediment texture, and geochemistry, with a chronology built using optically stimulated luminescence (OSL) dating. The data was contributed by the Australian Ocean Data Network and was last updated on 2026-04-04.
Canada's forested ecosystems, covering approximately 650 million hectares, are represented by this dataset of mean canopy height derived from lidar and Landsat data. The product was generated by Natural Resources Canada within the National Terrestrial Ecosystem Monitoring System framework for strategic-level forest monitoring from 1985 to 2011. It provides a 30-meter resolution raster layer for 2015, based on peer-reviewed science combining lidar plot metrics with Landsat pixel composites using a nearest neighbour imputation approach.
Canada's forested ecosystems, covering approximately 650 million hectares, are represented in this dataset of Lorey's mean height from 1985 to 2011. The data was generated by Natural Resources Canada using a nearest neighbour imputation approach that combines lidar plot information with Landsat pixel-based composites. Products are based on peer-reviewed science and have a spatial resolution of 30 meters.
1985–2011 time-series of forest canopy structure metrics for Canada's forested ecosystems, covering approximately 650 million hectares. The dataset was developed by Natural Resources Canada within the National Terrestrial Ecosystem Monitoring System (NTEMS) framework. It provides a 30-meter resolution raster layer representing the percentage of first lidar returns above the mean canopy height.
Canada's forested ecosystems, covering approximately 650 million hectares, are represented in this 30-meter resolution raster dataset. It provides the standard deviation of canopy height derived from 2015 lidar first returns, indicating variability in forest structure. The data was produced by Natural Resources Canada within the National Terrestrial Ecosystem Monitoring System framework, using methods described by Matasci et al. (2018).
Mudge Island and Link Island in British Columbia's southern Gulf Islands are the focus of this study on above-ground biomass (AGB) change from 2019 to 2024. Author Shangcheng Li combined PlanetScope imagery, Lidar point clouds, and stand-level inventory data to classify land cover and model AGB. The resulting biomass prediction model had an adjusted R² of 0.657 and was statistically significant (p = 0.005).
Macro-averaged segmentation performance on an external test set, comparing five baseline models to the proposed SPFF–UNet model. The dataset contains mean ± standard deviation metrics computed over 12 foreground classes, excluding background voxels, across three random seeds. This 5.5 KB Excel file was authored by Nadine Francis and last updated on April 9, 2026.
4.7 MB PDF contains in silico findings on transport and reaction mechanisms in electrochemotherapy. The work compares two pharmacokinetic profiles (TPK and UPK) across permutations of electric field magnitude (E) and bloodstream velocity (λinl). Simulations were conducted using a meshless code and a Boolean model to analyze concentration ratios.
A 5.5 MB PDF presents in silico findings from meshless and Boolean network simulations comparing transport and reaction mechanisms in electrochemotherapy. The study examines combinations of electric field magnitude (E), bloodstream velocity (λinl), and two pharmacokinetic profiles (TPK and UPK). It was authored by Fabián Mauricio Vélez Salazar and published in 2025.
3D LOD2 building models with textures cover several Montreal boroughs. The dataset includes CityGML, 3DM, and GDB formats, with DWG versions lacking textures for some areas. A digital terrain model is also available for a complete territorial representation.
Supplementary data on Hall measurements and texture coefficients of unspecified samples. The dataset was authored by Madhu S and is available under a CC-BY-4.0 license. It was last updated on 2026-05 04.
DeKH (German Hospital Dataset) is a multi-scene dataset of real hospital environments comprising high-resolution 3D point clouds with semantic annotations and ground-truth IFC BIM models. It was introduced by RPTU-FGMB alongside the BIMStruct3D pipeline and was accepted at EC3 2026. The dataset covers four distinct hospital scenes across three buildings.
Forest Percentage Above 2m 2015 is a dataset from Natural Resources Canada representing canopy cover. It provides percentage estimates of first returns above 2 meters for Canada's forested ecosystems, covering approximately 650 million hectares from 1985 to 2011. The data was generated by combining lidar plot information with Landsat pixel-based composites using a nearest neighbour imputation approach.
Rotterdam's buildings are modeled in 3D using the Dutch Key Registers of Addresses and Buildings (BAG) and Laser Altimetry data. Textures for the models are derived from aerial photography. The dataset is provided by the Dutch Ministry of the Interior and Kingdom Relations under a Creative Commons Public Domain Mark license.
362 participants across 9 cohorts from 8 clinical sites comprise the largest publicly available archive of 3D mesh gait data for Parkinson's Disease. The dataset was aggregated by the project team and is hosted on Hugging Face by the user 'vida-adl'. The dataset page was last updated on 2026-04-26.