Loading...
Loading...
Self-driving perception, LiDAR/camera fusion, trajectory prediction, drone perception, robot manipulation
1,662 datasets
A module from the Davis Logic V2 architecture, last updated June 2, 2026, by Jamie Davis. It provides a runtime-adaptive sliding-window filtering matrix and sensor fault isolation engine for real-time data streams. The module is implemented in freestanding, zero-heap C++11 and is designed for safety-critical processing in harsh environments.
TRANSSET is a dataset of 4,704 high-resolution (1024×1024) drone-captured images for traffic detection and vehicle classification. It contains 21,661 annotated vehicle instances across 8 classes, extracted from 4K video footage taken at two highway locations in North Carolina, USA. The dataset includes annotations in PASCAL VOC XML, YOLO TXT, and COCO JSON formats.
Spring 2024 data from three Mistbelt forests in the Amathole region, South Africa. The dataset comprises high-resolution RGB orthomosaics, 3D point clouds, derived vegetation indices, and ground-based leaf spectral data from 270 Podocarpus latifolius samples, collected to evaluate active forest restoration. It was created by researchers from the University of Fort Hare and submitted to GIScience & Remote Sensing.
The Cloud Physics Lidar (CPL) dataset provides airborne remote sensing measurements from the NASA ER-2 High Altitude Aircraft. It offers cloud and aerosol profiling with 30-meter vertical and 200-meter horizontal resolution at 1064 nm, 532 nm, and 355 nm wavelengths. Data products include aerosol and smoke plume profiling, optical depth estimates, and extinction profiles for analyzing radiative and optical properties of thin clouds.
July–December 2018 remotely sensed topographic and bathymetric data acquired for the New South Wales and southern Queensland coast using Airborne LiDAR Bathymetry. The dataset covers an area of 6862 km², providing combined terrestrial elevation and marine depth data at a 5 x 5 m horizontal resolution, processed to Australian Height Datum. Data were acquired by Fugro Pty Ltd on behalf of the NSW Office of Environment and Heritage and are provided via the Geosciences Australia ELVIS website.
112 ground-based lidar scans from the Echidna Validation Instrument capture forest structure across three U.S. campaigns. The dataset pairs these near-infrared canopy images with manual field measurements of tree species, DBH, height, and crown base. A validation file directly compares lidar-derived metrics like stem density and leaf area index to physical measurements from the 2007 New England campaign.
Iran experienced a magnitude 5.4 earthquake on August 05, 2025, with an epicenter 70km south of Mohammadabad-E Rigan. The dataset, produced by the WFP's Automated Disaster Analysis and Mapping (ADAM) system, provides geospatial and socio-economic impact analysis, including an estimate of 37,403 people affected within a 50km radius. It was last updated on May 21, 2026.
513,788 people were impacted within 50km of a magnitude 5.4 earthquake near Aliabad-E Katul on July 19, 2025. The dataset, provided by the World Food Programme's Automated Disaster Analysis and Mapping (ADAM) system, contains geospatial and socio-economic information for rapid humanitarian response. It was last updated on May 21, 2026.
Seabed landform features classified from the New South Wales statewide marine lidar dataset acquired in 2018. The dataset covers 4060 km2 of the NSW coast, extending from the shoreline to a 50-meter depth, and was created by the NSW Department of Climate Change, Energy, the Environment and Water using a publicly available ArcGIS toolset. It includes classified features such as reefs, plains, peaks, scarps, depressions, and channels.
Nan Zhang's research dataset contains performance metrics for a novel 3D object detection method evaluated on the KITTI and NuScenes benchmarks. The dataset, last updated in June 2026, includes segmentation and detection accuracy percentages, processing times, and frames-per-second metrics. The method integrates a Cloth Simulation Filter, an improved Euclidean clustering algorithm, and an enhanced PointNet architecture.
A study by Nan Zhang introduces a 3D object detection method for autonomous driving, evaluated on the KITTI and NuScenes benchmarks. The method integrates a Cloth Simulation Filter and an enhanced PointNet architecture, achieving segmentation accuracies of 94.96% and 93.12% with real-time processing speeds. The dataset, last updated in June 2026, contains the experimental setup and parameter configurations for this research.
Popat Shivaji Pawar's dataset compares UAV-based multispectral phenotyping against visual canopy wilting scores for evaluating soybean drought tolerance and yield retention. The data covers 85 genotypes from breeding programs in Arkansas, Missouri, Kansas, and North Carolina, tested under irrigated and rainfed conditions in Stuttgart, Arkansas from 2023 to 2025. UAV-derived vegetation indices showed significant correlations with yield, while visual scores showed weak associations.
85 soybean genotypes from multiple U.S. breeding programs were evaluated under irrigated and rainfed conditions from 2023 to 2025. Unmanned Aerial Vehicle (UAV)-based multispectral imagery captured vegetation indices at the R4/R5 growth stage, which showed significant correlations with yield under rainfed conditions. The dataset, authored by Popat Shivaji Pawar and shared under a CC-BY-4.0 license, compares these quantitative UAV metrics against traditional visual canopy wilting scores.
A dataset of 85 soybean genotypes evaluated under irrigated and rainfed conditions in Stuttgart, Arkansas from 2023 to 2025. It compares visual canopy wilting scores with UAV-derived multispectral vegetation indices and their correlation with grain yield. The data was authored by Popat Shivaji Pawar and shared under a CC-BY-4.0 license.
Crown-BERT achieved overall accuracies of 83.1% to 90.8% on three UAV datasets for classifying individual tree species. The lightweight deep learning framework, proposed by Xin-Bo Wang, uses fused hyperspectral and LiDAR data and was last updated in May 2026. It introduces crown-morphology-aware techniques like dynamic crown masking and self-supervised pre-training to improve accuracy and reduce manual annotation.
A 22.2 KB dataset quantifying transverse oscillations of the soleus muscle in response to impact forces during human locomotion. The data, shared by Emma Tole under a CC-0 license, was collected using 2D ultrasound imaging during walking and running at various speeds. It was last updated on 2026-05-28.
A 5.5 KB Excel file containing Denavit-Hartenberg (D-H) parameters for a 6-degree-of-freedom Cartesian coordinate robotic arm used in mine pipeline installation. The dataset supports a study proposing an improved trajectory planning algorithm (AS-DTRRT) to reduce runtime by up to 20.22% and path length by up to 21.17% compared to other methods. Author Zhixiang Liu uploaded it to figshare under a CC-BY-4.0 license, with a last update timestamp of 2026-05-22.
New England region data provides 30-meter gridded estimates of aboveground biomass density, forest canopy height, and tree canopy coverage for the nominal year 2015. The dataset, produced by NASA, integrates high-resolution leaf-off LiDAR data from 2010-2015, agricultural imagery, and U.S. Forest Service plot measurements. It supports the development of a cohesive forest carbon accounting system at national and international scales.
IDMC's Internal Displacement Updates (IDU) dataset provides provisional daily data on new internal displacements (flows) in Iran caused by conflict and disaster. The data is sourced from initial assessments and updated daily, representing the most recent available information. Finalized and validated estimates are later published in the annual Global Internal Displacement Database (GIDD).
2021 saw a complete airborne LiDAR survey of the Northern Ireland coastline, commissioned as part of the NI 3D Coastal Survey. The resulting Digital Terrain Model provides precise data on current coastal morphology, covering the intertidal area and extending approximately 200 meters landward of the high-water mark. This dataset is available on multiple open data platforms under a permissive license.