Loading...
Loading...
Image classification, object detection, segmentation, face recognition, OCR, image generation, video understanding
16,995 datasets
Image and text data from Wikipedia Simple English, with crowdsourced annotations. The dataset size is categorized as between 100K and 1M instances, and it is stored in Parquet format.
27,000 synchronized frames of 16-layer LiDAR, RGB camera, and 4D radar data were collected across multiple buildings at the University of British Columbia. The Indoor FireRescue Rada (IFR) dataset includes raw ADC data, point clouds, and 3D bounding box annotations for objects like doors, chairs, and fire hydrants. This dataset was created by author yysd123 and last updated on Hugging Face in October 2025.
From 1978 to 2004, this dataset provides a view of the consumption expenditures of private households in the Netherlands, detailing their size, composition, and financing. It is a protected microdata file from Statistics Netherlands (CBS), processed to prevent the identification of individuals or households. The data is hosted by the DANS Data Station Social Sciences and Humanities Collection.
Dutch data on granted construction permits for residential and non-residential buildings. The dataset is a protected microfile from the Central Bureau for Statistics (CBS), processed to prevent identification of individuals or households. It covers annual data from 1987 through 2009, excluding 1997.
The U.S. Environmental Protection Agency's Automotive Trends Report provides data on greenhouse gas emissions, fuel economy, and technology for every new light-duty vehicle model sold in the United States since 1975. The data is collected from EPA laboratory testing and manufacturer submissions to support regulatory programs like GHG and CAFE standards. It is available in annual reports and downloadable spreadsheet formats.
A lookup table mapping civil parishes and communities to electoral wards and local authority districts in England and Wales as of 31 December 2021. The dataset includes codes and names for each administrative level and was published by the Government Digital Service. It was last updated on the platform on 11 August 2025.
Entrevistas exposición Zhongguo is a case study within the R&D project 'Memory, Forgetting and Crossing Borders in East Asia' directed by Prof. David Mervart and Taciana Fisac. The dataset contains two semi-structured interviews conducted in May 2024 with participants involved in organizing the 'Zhongguo: el País del Centro' exhibition at the National Museum of Anthropology. The study aims to analyze the contribution of descendants of Chinese immigrants in Spain to the construction of the historical memory of the Chinese community.
Bianca Vitalaru's qualitative dataset, harvested via e-cienciaDatos from Dataverse, collects keywords and brief ideas from participants on translation and interpretation in public services. The data, last updated on October 14, 2025, covers participant profiles, hiring requirements, and sought competencies. Information is grouped by sector, including public institutions, NGOs, private companies, and freelance professionals.
Developed by KlingTeam for SIGGRAPH Asia 2025, this dataset supports scene-consistent interactive long video generation through memory retrieval. It is distributed as a multi-part archive containing video sequences designed for maintaining visual coherence over extended durations.
32 manually digitized shoreline shapefiles capture the coastline near Masset, British Columbia, twice annually in April and October. Madison Power created this dataset to investigate shoreline position changes over a 15-year period using the U.S. Geological Survey's Digital Shoreline Analysis System (DSAS). The seasonal shorelines were derived from Landsat 7 and 8 panchromatic imagery, using the water-land interface as a proxy.
National Statistics Postcode Lookup (NSPL) for the United Kingdom as of February 2021 relates current and terminated postcodes to statutory geographies via best-fit allocation from 2011 Census Output Areas. The 196 MB file is produced by ONS Geography and issued quarterly to support area-based statistics from postcoded data.
MemoryAgentBench is a benchmark for evaluating LLM agent memory and retrieval capabilities, released by ai-hyz in July 2025. It contains fewer than 1,000 samples across specialized tasks including DetectiveQA, Longmemeval, and LRU-based keypoint tracking. The dataset was iteratively refined through October 2025 to remove high-cost and inefficient evaluation samples.
Ok Viwikitext Image Captioning is a dataset published on HuggingFace by NezuMiii. The title suggests it contains image-text pairs, likely for training or evaluating multimodal AI models. The dataset was last updated on November 19, 2025.
Openimage-O is an out-of-distribution benchmark subset derived from the Open Images dataset. It originates from the OpenOOD benchmark evaluation splits. The dataset was uploaded by the user 'torch-uncertainty' and was last updated on October 10, 2025.
Mimic Yolo Prediction Cxr likely contains chest X-ray images annotated for object detection tasks. The dataset was published by author Rookiezz on HuggingFace and was last updated on November 14, 2025. Its specific size, annotation details, and exact content require verification after download.
CHIRLA provides approximately 1 million identity-annotated bounding boxes across 596,000 frames for long-term person re-identification and tracking. Captured by bdager over a 7-month period, the data utilizes 7 cameras within connected indoor environments to track 22 unique identities.
Grace Jeanpierre's dataset from the Texas Data Repository provides mechanical data from rhesus monkey brain tissue. It includes 35 samples from 4 subjects, covering both basal ganglia (gray matter) and corona radiata (white matter). Four types of mechanical tests were performed on a triaxial testing machine: compression relaxation, tension relaxation, tension-compression cycling, and shear cycling.
Images of acacia trees and non-acacias, along with corresponding segmentation masks, hosted on Hugging Face by author duartepcruz. The dataset is structured with separate folders for RGB images and segmentation masks for test and validation splits. It was last updated on 2025-10-29.
OmniDocBench contains 1,355 PDF pages across 9 document types and 3 languages, developed by OpenDataLab for evaluating document parsing systems. The dataset provides 15 block-level layout annotations for diverse materials including academic papers, financial reports, and handwritten notes.
10 MRI studies from different angles provide labeled scans of brain cancer. The dataset includes DICOM files and accompanying PDF reports annotated by doctors. UniqueData created this collection, last updated in September 2025.