Loading...
Loading...
3D models, rendered datasets, physics simulation, digital twins, synthetic data generation, game engine data
1,411 datasets
18,074 samples of steel surface imagery annotated for defect detection, curated by Voxel51 and hosted on Hugging Face. The dataset is designed for computer vision tasks in industrial manufacturing quality assurance. It was last updated on May 19, 2026.
OmniGameArena is a benchmark for evaluating game-playing agents across a diverse suite of interactive video games. The dataset provides packaged, ready-to-run game environments built with Unreal Engine 5, authored by mxlin043. It was last updated on June 2, 2026.
Geoscience Australia provides Web-viewable 3D models integrating coastal spatial data for the Keppel Bay and Fitzroy River area in Queensland, Australia. The models incorporate digital elevation models, multibeam bathymetry, sediment samples, benthic habitats, and satellite imagery. These models are designed for sharing data with project partners and coastal zone stakeholders via an easy-to-use web interface.
Synthetic CAD construction videos are paired with per-frame atomic action operations for training video-to-action and image-to-video world models. Version 3 extends the v2 dataset with batches b07 through b14, maintaining an identical format. The dataset was created by author hz6666 and was last updated on June 4, 2026.
Australian Ocean Data Network hosts a dataset detailing a 1500-year record of coastal sediment accumulation preserved in beach deposits at Keppel Bay, Queensland, Australia. The study examines ridge morphology, sediment texture, and geochemistry, building a chronology using optically stimulated luminescence dating. The dataset was last updated on 2026-05-05.
3DReflecNet is a large-scale synthetic multi-view dataset for novel view synthesis and 3D reconstruction. It provides Blender-rendered RGB images, masks, depth maps, normal maps, camera parameters, and 3D model assets. The dataset was created by 3DReflecNet and last updated on Hugging Face in June 2026.
Preprocessed object meshes packaged for use with the SimVLA project. The dataset contains approximately 37 GB of data across 2,587 object directories. It is a re-distribution of meshes generated by the BODex preprocessing pipeline and was uploaded by exaFLOPs09 on Hugging Face.
Monthly (or 3-monthly at some sites) biomass sampling of zooplankton communities at nine stations around the Australian coastline, with some data from Port Hacking dating back to 2002. The dataset, part of the Integrated Marine Observing System (IMOS), includes community composition, dry weight biomass, and size spectrum analysis. Data is managed by the Australian Ocean Data Network and is interoperable with other national and international programs.
Njc Kitchen is a dataset of everyday kitchen objects built by notjustchairs.ai. It contains a target of approximately 1,000 verified 3D assets across six categories, including appliances, cookware, and utensils. The dataset was last updated on June 5, 2026.
FOR-instanceV2 is a dataset of forest LiDAR 3D point clouds used for end-to-end segmentation. It was created by author zhan994 for the paper "ForestFormer3D: A Unified Framework for End-to-End Segmentation of Forest LiDAR 3D Point Clouds". The dataset was last updated on the Hugging Face platform on June 13, 2026.
The Australian marine region, covering the extent of the AusBathyTopo 250m 2023 grid, is represented by this dataset. Geoscience Australia collated sediment properties for over 15,000 seabed samples from its Marine Sediment Database (MARS), an online resource of approximately 2.6 million entries. Analytical data includes textural composition, particle size statistics, textural class, and calcium carbonate values, with sample location, water depth, and survey information.
Harpreetsahota's Graspclutter6D dataset is a FiftyOne dataset with 111 samples. The dataset page on Hugging Face provides further details. Its last recorded update was on 2026-05-21.
July 2003 data provides soil texture classifications for Soil Climate and Analysis (SCAN) sites in northern Alabama, USA, as part of the Soil Moisture Experiment 2003 (SMEX03). This ancillary dataset supports the analysis of soil moisture by characterizing the soil composition at specific monitoring locations. The data is associated with NASA and the National Snow and Ice Data Center (NSIDC).
Pixel Art Normal Maps is a synthetic dataset of image pairs for training and evaluating normal-map generation models for 2D pixel art. It accompanies a master's thesis on normal-map generation for pixelated images. The dataset was created by DmytroKhitro and is hosted on Hugging Face.
WithinUsAI created the Nikola Tesla Mindstate 25K Dataset, a high-quality synthetic dataset of 25,000 unique training examples. It was last updated on 2026-05-25. The dataset is designed to fine-tune any LLM to embody Nikola Tesla's mindset, thinking patterns, and innovative methodology.
Per-object 3D manipulation extractions from the EPIC-KITCHENS-100 dataset. The data was used to train ObjectForesight, a 3D object-centric dynamics model that predicts future 6-DoF object poses and trajectories from short egocentric video. The dataset was authored by raivn and last updated on June 6, 2026.
A longitudinal and experimental dataset from a randomized controlled trial investigating the relationship between future self-identification components and goal achievement. The study involved 89.4% female first-year university students with a mean age of 19.59 years, assigned to a control condition or one of two FutureU intervention conditions. Data was collected by author Tettero, Tiffany and last updated on the platform in May 2026.
June 2009 biological and acoustic data collected during a five-day transit from Nelson, New Zealand to Devonport, Australia. The dataset was gathered by CSIRO Marine and Atmospheric Research using a MIDOC net system, a 38 kHz echosounder, and an Acoustic-Optical System attached to commercial trawl nets. It likely contains information on micronekton distribution and acoustic reflectivity from the surface down to 1500 meters depth.
WildCity is a dataset from The University of Tokyo designed to train feed-forward 3D Gaussian Splatting models for 3D reconstruction from unconstrained photo collections. It comprises multi-view, multi-lighting images covering 200 scenes and 170 HDRI maps, and includes diverse transient objects. The dataset was last updated on June 11, 2026.
A historical depth map of the IJsselmeer area created by the Dienst der Zuiderzeewerken. The map likely dates from around 1935 and depicts the Wieringermeer reclamation, the Noordoostpolder, and the Afsluitdijk. The Cultural Heritage Agency of the Netherlands holds the original analogue map, and the provided scan is georeferenced to the Dutch National Triangle System (RD).