Loading...
Loading...
Image classification, object detection, segmentation, face recognition, OCR, image generation, video understanding
16,126 datasets
The Moltbook Social Interaction Dataset captures large-scale social interactions from Moltbook, a persistent online platform populated entirely by LLM-powered agents. The dataset was created by AIcell and last updated on February 18, 2026. It records autonomous agent activities such as creating posts, writing comments, voting on content, and engaging in threaded discussions across topic-based groups called 'submolts'.
A dataset derived from the HAM10000 collection of dermatoscopic images. It likely contains synthetic images generated using SMOTE and GAN techniques for data augmentation. The dataset is hosted on Kaggle, but its specific creation date and author are unknown.
This dataset documents 12 specific failure cases identified in the Qwen3.5-4B-Base multimodal model. Created by PhoenixA and updated in March 2026, it categorizes blind spots across 10 domains including arithmetic, negation, and visual grounding. The data focuses on models utilizing Gated DeltaNet and sparse Mixture of Experts (MoE) architectures.
London Borough of Camden maintains a register of licensed Houses in Multiple Occupation (HMOs) and their management arrangements. The dataset includes details such as the maximum permitted number of persons, number of bedrooms, and amenities, as populated from license applications. It is intended for identification purposes, not marketing, and was last updated in March 2026.
Featuring dendrochronological data from wooden samples taken from three ancient water wells discovered during an archaeological survey in Koekelare. The analysis includes samples from oak (Quercus spp.) well casings, with specific dating results for wells labeled spoornr. 9013 and spoornr. 10048. One well was dated to the transition between the 2nd and 3rd centuries BC, while another was dated to the early 12th century AD.
The U.S. Geological Survey's data release documents germanium production sites, with public records for a single U.S. mine. It details the Apex mine in Utah, which produced germanium in the 1980s, and notes other potential byproduct sources in Alaska, Washington, and Tennessee. The dataset compiles information on deposits, production history, and critical mineral classifications.
A dataset titled 'hengck23-rogii-cnn-mtp-demo' published on Kaggle. The title suggests it is likely a demonstration dataset for a Convolutional Neural Network (CNN) model, possibly related to a project or user named 'hengck23' and 'rogii'. Its specific content, size, and creation details are not provided in the available metadata.
Person_GAN is a dataset hosted on Kaggle. The title suggests it contains images generated by a Generative Adversarial Network focused on human figures. No further metadata regarding size, authorship, or creation date is available.
A dataset titled 'models (cnn_scratch)' published on Kaggle. The title suggests it contains data related to a convolutional neural network built from scratch, likely including model weights or architecture definitions. No further metadata is available to confirm its specific contents, size, or origin.
An OCR dataset for the Moroccan Arabic dialect, Darija, published on Kaggle. The dataset likely contains images of handwritten or printed text paired with transcriptions for optical character recognition tasks. Its specific size, creation date, and author are unknown.
A collection of convolutional neural network models shared on Kaggle. The dataset likely contains model architectures and weights trained from scratch on an unspecified task. Its specific contents, such as the number of models or their training data, require verification after download.
A dataset published on Kaggle for object detection tasks related to autonomous vehicles. It likely contains images annotated for use with the YOLOv8 model. The specific source, size, and creation date are unknown.
A collection of medical images contributed by an individual named Lavanya and hosted on Kaggle. The dataset's specific content, scale, and creation date are not detailed in the provided metadata. Its intended application appears to be within the medical imaging and computer vision domain.
A dataset likely containing synthetic retinal images generated using the PGGan and Pix2PixHD models. The specific volume, source, and creation date are unknown. It is hosted on the Kaggle platform.
Kaggle hosts a dataset titled PGD_PNG_ATTACK_PER_SEGMENT_NOISE_GAN. The title suggests it contains images generated for adversarial attacks, likely using Projected Gradient Descent (PGD) and Generative Adversarial Networks (GANs). Metadata such as author, size, and license is unknown.
MIT-BIH_SPLIT_GAN likely contains electrocardiogram (ECG) time-series data derived from the well-known MIT-BIH Arrhythmia Database. The title suggests the data has been processed or split specifically for training Generative Adversarial Networks (GANs). Published on Kaggle, the dataset's exact size, format, and specific modifications require verification after download.
8,181 high-quality synthetic invoice images for OCR and document understanding tasks, created by Voxel51. The dataset includes 1,489 fully annotated samples with structured JSON metadata and raw OCR text, plus 6,692 unannotated images.
A collection of over 12,000 images spanning multiple categories. The dataset is multimodal, which likely indicates it contains images paired with other data types like text or labels. It was published on Kaggle, but details about the author, organization, and specific creation date are unknown.
junyeong-nero's synthetic OCR dataset was automatically generated on 2026-02 19 and uploaded to Hugging Face. The dataset is described as containing Chinese (zh) text images created by a synthetic generation pipeline. The specific scale and structure of the data are not detailed in the provided metadata.
FDGAN-Code is a dataset published on Kaggle. Its title suggests a focus on code related to Generative Adversarial Networks (GANs). The dataset's specific content, scale, authorship, and recency are unknown from the provided metadata.