Loading...
Loading...
Image classification, object detection, segmentation, face recognition, OCR, image generation, video understanding
14,846 datasets
Jiaqi Li curated a collection of 22 commercial video advertisements for consumer display technologies. The corpus spans from 1960 to 2026 and includes brands like Sony, Samsung, LG, and RCA. It was assembled from YouTube and is intended for non-commercial academic research.
A 2026 figshare dataset by Astrid Dagmar Bernkop-Schnürch investigates chlorido iron(III) complexes. The dataset likely contains characterization and activity data for complexes differing in stereochemical configuration and halogen substituents. It focuses on the relationship between molecular structure and antitumor efficacy.
Moses Akujobi created a 1.1 GB dataset for vehicle detection, tracking, and counting from roadside video. It includes a dedicated class for keke (auto-rickshaw/tricycle), which is not a standard class in common international benchmarks. The dataset was last updated on April 9, 2026 and includes train/validation/test splits with YOLO-format labels.
34 skills for the Chrome domain and 26 for GIMP are included in this dataset from the Towards MMSkills project. It contains procedure descriptions, runtime state cards, and visual references organized by application domain. The dataset was authored by zhangkangning and last updated on 2026-05-11.
Vaskor Mostafa's dataset compares the LeafDet model against the YOLOv8n baseline. The dataset is a 5.5 KB Excel file, last updated on May 22, 2026. It is shared under a CC-BY-4.0 license on the figshare platform.
A figshare dataset by Rajasekar Marimuthu, last updated in April 2026, presents crystallographic information for synthetic compounds. The data supports a published method for generating chelated ε-N-Ts-amidoallylindiums from chiral allylic alcohols, which undergo stereocontrolled α-allylation of aldehydes. The 597.1 KB file contains structural data for the resulting homoallylic alcohols and derivatives.
Zijian Chen published a performance comparison of different attention modules used with the YOLOv11n baseline model for object detection. The dataset, last updated on 2026-05-22, is a 9.5 KB Excel file containing evaluation results based on the Tomato-Village dataset.
Seabed mapping and habitat classification surveys completed in Darwin Harbour during 2011 and 2013. The research was a collaboration between Geoscience Australia, the Australian Institute of Marine Science, the Department of Land Resource Management, and the Darwin Port Corporation. Key outcomes include detailed bathymetry maps and a seascape map with six habitat classes derived from substrate, relief, and bedform data.
A reinforcement-learning environment from the Poolside Laguna Hackathon submission teaches an LLM to reason like a computational chemist. The dataset is part of a tool-use environment focused on measuring protein-ligand interactions rather than guessing. It was created by Team JAMMY and last updated on May 31, 2026.
Parameter settings for a proposed CNN-SVM model, including details on CNN architecture and optimization choices. The dataset was authored by Haiyan Lu and last updated on May 22, 2026. It is a 5.5 KB Excel file.
A 9.5 KB Excel file containing data on socio-demographic and organizational factors linked to workplace violence. The dataset was authored by Nwanneka Chidinma Ghasi and last updated on May 22, 2026. Its specific temporal coverage and row count are not detailed in the provided metadata.
Ablation experiment results on the PASCAL VOC2012 dataset are provided in a 5.5 KB XLS file. The dataset was authored by Haiyan Zhang and last updated on May 22, 2026. It is shared under a CC-BY-4.0 license on the figshare platform.
Ocean Drilling Program Leg 189 recovered 4539 meters of marine core from five sites in the Tasmanian Gateway, with a recovery rate of 89%. The dataset, hosted by the Australian Ocean Data Network, documents upper Maastrichtian to Holocene sediments to study the opening of the gateway and its impact on the Antarctic Circumpolar Current and global climate. The sedimentary sequence includes shallow marine mudstones, glauconitic siltstones, and pelagic carbonates.
Northern Lord Howe Rise submarine volcanic cones were mapped and dredged at 23-24°S, about 120 km apart. Dredged rocks include altered hyaloclastites and basalt, with interbedded micrites dated to the late Early Miocene (~16 Ma). Ferromanganese crusts up to 7 cm thick were sampled, with chemical analyses showing variations between the northern and southern cones.
Image-to-text OCR pairs extracted from a Hassaniya Arabic stories corpus. Each row couples an image crop from a source book with its manually cleaned text. The original material, titled '41 Short Stories About Life in Mauritania, Especially Life in Nouakchott', was published in 1994 and written by K. ould Beye and Dave Penney.
GURAS is a 'property' based address database serving as a single source of truth for New South Wales. Each property polygon has a unique numeric identifier and contains at least one authoritative address sourced from local councils via the valuation of land database. The dataset is managed by Spatial Services (DCS) and was last updated on 2026-04 09.
8 document images were processed for text recognition using the GLM-OCR model on 2026-06-05. The dataset contains OCR results generated from the source dataset davanstrien/ocr-affordances-pages. Processing was completed in 2.9 minutes by author davanstrien.
A mixed-method study explores knowledge, opportunities, and challenges for primary school adolescents with disabilities handling menstruation in Mbarara district, Southwestern Uganda. The dataset, shared by sserunkuuma Jonathan, is licensed CC-BY-4.0 and was last updated in May 2026. Its small size of 3.4 KB suggests it likely contains qualitative and quantitative survey data.
A cross-classification table showing the distribution of relationships between OHPs and organizations. The table, created by Sheila Timp and last updated in May 2026, is a 9.5 KB Excel file. It counts pairs where rows indicate the number of organizations per OHP and columns indicate the number of OHPs per organization.
NASA HEASARC provides a catalog of gamma-ray observations from the INTEGRAL satellite, which began collecting data in October 2002. The catalog is a collaborative effort between the INTEGRAL Science Data Center in Switzerland and NASA Goddard Space Flight Center, based on data from the IBIS and SPI instruments. It was first created in September 2004 and was updated weekly until June 2019.