Loading...
Loading...
Image classification, object detection, segmentation, face recognition, OCR, image generation, video understanding
15,958 datasets
GIS files contain a base-square grid with distance and surface analysis attributes for each cell, supporting user-defined queries. The data includes aerial imagery classified into natural and non-natural areas for parks within the Southeast Coast Network. Procedures are designed for analysis in the FRAGSTATS landscape application and adaptation for spatially-balanced sampling.
Senthil12345 provides a dataset of 445,213 daily weather records for Telangana, India, sourced from Open Data Telangana. It covers the period from February 1, 2023, to January 31, 2025, with district and mandal-level granularity. The data includes fields such as rainfall, temperature, wind speed, and humidity.
Model performances for the OctMNIST dataset, evaluated using the VGG-16 architecture. The dataset contains test accuracy, average inference time, and total inference energy consumption, expressed as relative percentual differences to benchmark values. It was authored by Amirhossein Douzanneh Zenoozi and last updated on March 18, 2026.
Relative performance metrics for deep learning models on two standard computer vision datasets. The data includes test accuracy, inference time, and energy consumption, expressed as percentual differences from a benchmark. Author Amirhossein Douzandeh Zenoozi published the 5.5 KB Excel file on figshare in March 2026.
Seth H. Frisbie's dataset provides daily manganese (Mn) exposure estimates in milligrams per day (mg/day) for children from 6 months to 5 years old. The data is based on measured Mn content in products, with separate estimates for each age covered by product label indications. It was last updated on March 18, 2026, and is shared under a CC-BY-4.0 license.
A dataset likely containing images of residential fires intended for training YOLO-based computer vision models. It was published on Kaggle, but its size, creation date, and author are unknown. The specific content and annotations require verification after download.
A dataset likely containing images for training object detection models to identify smoke and fire. It is hosted on Kaggle, but specific details like size, source, and annotation quality are not provided. The dataset's creation date, author, and exact composition are unknown.
MA-EgoQA features 266 hours of multi-agent egocentric video and 1.7k question-answering pairs, developed by Kangsan Kim in 2026. The collection captures six individuals living together for seven days, providing a benchmark for simultaneous long-horizon video understanding.
A collection of CT scan images labeled for binary classification of lung cancer. The dataset is hosted on Kaggle and is intended for medical image analysis tasks. The author, organization, and specific collection details are not provided.
A 2026 publication from Geoscience Australia examines organic geochemical studies of reduced organic material in sedimentary rocks. The research provides insight into sediment origin and history by analyzing biological markers to determine organic matter source, depositional environment, biodegradation, and maturity. The dataset is presented in HTML and PDF formats.
Loongana SH52-09 is a digital topographic map from the AUSTopo series covering Australia at a 1:250,000 scale. The series comprises 516 maps, the largest scale providing full continental coverage, with each standard map depicting an area of about 1.5 by 1 degrees. It contains natural and constructed features including road and rail infrastructure, vegetation, hydrography, and 50-meter contours.
KazakhOCR is a synthetic benchmark dataset for evaluating multimodal models on Optical Character Recognition (OCR) for the Kazakh language. It contains text in Arabic, Cyrillic, and Latin scripts. The dataset was curated by Henry Gagnier, Sophie Gagnier, and Ashwin Kirubakaran and is licensed under MIT.
Maharashtra, India, is the geographic focus of this curated image dataset of mango leaves. The dataset likely contains images from major regional mango varieties, though the exact number of images and specific varieties are not detailed. The author, organization, and collection date are unknown.
Eleven years of dropsonde observations from 1950 to 1961 (excluding 1953) were collected by the U.S. Air Force along the Ptarmigan reconnaissance track. The data contains vertical profiles of geopotential height, temperature, and dew point temperature, primarily for the lower troposphere below 500mb. The observations were made mainly over the Beaufort Sea and the western Arctic Ocean.
National Ocean Service (NOS) Coastal Wave Program data from 1979 to 1983, organized by NOAA NCEI. The dataset contains analyzed wave information structured into three fixed-length ASCII record types describing environmental conditions, wave energy spectra, and waverider measurements. All records are 128 characters long and include unique file type and identifier columns.
Australia's Gilgandra region is covered by this 1:250,000 scale digital topographic map from the national AUSTopo series. The series comprises 516 standard maps covering the entire continent at its largest published scale, with each sheet depicting approximately 150 by 110 kilometers. It contains natural and constructed features including infrastructure, vegetation, hydrography, and 50-meter contours.
Sampling for vertical plankton migrants was conducted over 24-hour periods in Edisto Inlet during 1968-69. The study, by SCIOPS, aimed to assess low light effects under ice but reported a general account of captured organisms due to strong tidal currents. Hauls were taken through a seal hole in the ice using a 50cm diameter net with 250µm mesh.
Replication Data for 'Disentangling the Three Facets of Mass Ideological Polarization' provides network-based measures of ideological disagreement and alignment. The dataset uses public opinion data from the World Values Survey and European Values Study, covering 78 societies between 2017 and 2023. It was authored by Guo, Yufan and published by Public Opinion Quarterly.
OCR Stress Test v2 is a multilingual benchmark dataset for evaluating optical character recognition systems. It is hosted on Kaggle, but detailed metadata about its size, structure, and creation is unavailable. The dataset likely contains images with text in multiple languages designed to test OCR robustness under challenging conditions.
UAV-based Vine Leaf Instance Segmentation likely contains aerial imagery of vineyards captured from an altitude of 20 meters. The dataset is designed for instance segmentation tasks, focusing on individual vine leaves. The author, organization, and last update date are unknown.