Loading...
Loading...
Image classification, object detection, segmentation, face recognition, OCR, image generation, video understanding
16,086 datasets
Surya OCR is a dataset for optical character recognition tasks, published on Kaggle. The dataset likely contains images of documents and corresponding text annotations. Specific details on size, source, and creation date are unavailable from the provided metadata.
A dataset for object detection focused on three culinary herbs: parsley, coriander, and epazote. The description indicates it contains bounding box annotations for these plants. The dataset's author, organization, and other metadata are unknown.
A sample dataset of dried chilli images intended for computer vision tasks. The dataset likely contains images for classifying different chilli varieties and assessing their quality grades. Its author, organization, size, and temporal coverage are unknown.
CMRAG-Bench is a test dataset created for the CMRAG paper. All PDFs have been rendered and parsed by the Qwen2.5-VL-72B model. The dataset includes rendered page images, parsed HTML content, and JSON files containing sub-image bounding boxes.
FANVID is a benchmark dataset designed for face detection, face matching, and license plate recognition under challenging low-resolution surveillance video conditions. The dataset, created by kv1388, features faces and license plates that are unrecognizable in individual frames, encouraging models to leverage temporal context across video sequences. The dataset page was last updated on March 1, 2026.
MOSAIC aggregates individual functional magnetic resonance imaging datasets using a shared preprocessing and stimulus curation pipeline. The dataset is designed to provide the scale needed for neural network training and the diversity required for generalizable results. It is hosted on AWS Open Data and is licensed under CC-BY-4.0.
Multi-temporal aerial imagery of rivers in the Philippines, 2019-2020 contains orthorectified aerial imagery for two river segments. The data covers the Bislak River downstream and the Abuan, Bintacan, and Pinacanauan de Ilagan Rivers confluence, captured during repeat surveys in 2019 and 2020. The resulting orthoimagery has a 0.2-meter spatial resolution with RGB bands.
Eight Welsh upland river reaches were sampled monthly from December 2012 to April 2013 to examine suspended organic matter stocks. The experiment contrasted moorland and conifer forest sites, with half the reaches receiving deciduous leaf additions and half serving as controls. Data collection was organized by Dr Isabelle Durance under the NERC-funded DURESS project.
41 rivers across Great Britain were sampled monthly throughout 2017 for the LOCATE project. The dataset includes concentrations of particulate and dissolved organic carbon, nutrients like ammonia and nitrates, alkalinity, pH, and isotopic signatures. It was produced by a multidisciplinary NERC consortium involving the National Oceanography Centre and British Geological Survey.
Predictions from a Convolutional Neural Network (CNN) model for the BirdCLEF 2026 competition. The dataset likely contains model outputs for identifying bird species from audio recordings. It is hosted on Kaggle, a platform for data science competitions.
Garbage Classification Dataset is a collection of images intended for training deep learning and CNN models. The dataset's specific size, source, and creation details are not provided in the available metadata. Its primary purpose is to facilitate automated waste sorting through computer vision.
City of Austin records identify food-permitted businesses required to implement organics diversion programs under the Universal Recycling Ordinance. The dataset includes properties phased in by size thresholds starting October 1, 2016, supporting the city's Zero Waste goal. Row and column counts are not specified in the available metadata.
A collection of brain MRI images likely processed for skull-stripping, a key step in neuroimaging analysis. The dataset is hosted on Kaggle, but its exact size, origin, and creation date are unspecified. It appears designed for training or evaluating convolutional neural networks for automated brain extraction.
fire-smoke-yolo-dataset is a collection of images for training object detection models. The dataset likely contains annotated images of fire and smoke, intended for use with the YOLO (You Only Look Once) framework. It is hosted on Kaggle, but details on its size, origin, and creation date are unknown.
Pantanal-EfficientNet-Weights is a dataset of pre-trained neural network weights published on Kaggle. The title suggests it is based on the EfficientNet architecture, likely intended for image classification tasks. The dataset's specific content, size, and creation details are not provided in the available metadata.
PaddleOCR is a dataset hosted on Kaggle. The dataset likely contains images and corresponding text annotations for optical character recognition tasks. Metadata such as row count, file formats, and license details are not provided in the input.
PPOCR_v5 is a dataset published on Kaggle. The title suggests it is related to optical character recognition, likely containing images and text for training or benchmarking models. Metadata is minimal; actual content, scale, and authorship require verification after download.
YOLOv12 source code is hosted on Kaggle. The dataset likely contains the implementation files for the YOLOv12 object detection model. The author, organization, and last update date are unknown.
HuBMAP YOLOv12 Weights is a dataset of model weights for a YOLO-based object detection model. It is hosted on Kaggle and is associated with platform tags for image data and computer vision. The dataset likely contains the trained parameters for a YOLOv12 model, potentially for use in biomedical image analysis related to the HuBMAP initiative.
An experimental dataset titled 'exp31_sage_morgan_5feat' published on Kaggle. The title suggests it likely contains five engineered features for a machine learning task. Its specific content, authorship, and scale require verification after download.