datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
dfhq-datasetKandenAiHackathonPosture2
KandenAiHackathonPosture2 — エンジニア姿勢合成画像データセット
English Summary
A large-scale synthetic image dataset of 52,500 images depicting engineers in four sitting postures (good, slouch, chin rest, stretch). Generated using Stable Diffusion v1.5 + ControlNet (Depth) from reference base images, with diverse variations in age, ethnicity, clothing, lighting, and camera artifacts. Designed for training posture classification models for real-time fatigue detection in office environments.… See the full description on the dataset page: https://huggingface.co/datasets/SeiyaCM/KandenAiHackathonPosture2.Alzheimers-Classification-Dataset
Dataset Card for Alzheimer's Classification Dataset
Dataset Summary
This dataset is an ML-ready derivative of the OASIS (Open Access Series of Imaging Studies) brain MRI database, curated and preprocessed for Alzheimer's disease stage classification. It contains 10,880 MRI brain scan images split across training, validation, and test sets, each labeled with one of four dementia severity stages.
The OASIS project is a collaborative initiative aimed at making brain… See the full description on the dataset page: https://huggingface.co/datasets/kanaadlimaye/Alzheimers-Classification-Dataset.kanops-open-retail-imagery
Kanops — Open Access · Imagery (Retail Scenes v0)
The first longitudinal retail imagery dataset for computer vision.
Every existing retail CV dataset (SKU-110K, RP2K, Freiburg Groceries, MVTec D2S) captures a single moment in time. Kanops captures 15 years of change — how stores, shelves, packaging, and merchandising actually evolve.
This open sample contains ~10–11K retail imagery files from UK and US stores (2010–2024). The full Kanops archive spans 1.135M+ retail images across… See the full description on the dataset page: https://huggingface.co/datasets/dresserman/kanops-open-retail-imagery.Sap_Kush_Med_Deepfake
Sap_Kush_Med_Deepfake Dataset
Paired medical-image forgery lineages across six modalities. Every lineage is
one source image, one mask, one seed: the arms differ only in what was done
inside the mask, so a comparison between arms isolates the manipulation rather
than an encoding artefact.
3956 lineages, 30385 files, 8.49 GiB.
v2 adds a removal arm grounded in human annotation for three more modalities
(endoscopy, ultrasound, MRI). v1 had removal for CT only.
What the… See the full description on the dataset page: https://huggingface.co/datasets/Kanhaiyya/Sap_Kush_Med_Deepfake.Tinul
Dataset Card for tiny-imagenet
Dataset Summary
Tiny ImageNet contains 100000 images of 200 classes (500 for each class) downsized to 64×64 colored images. Each class has 500 training images, 50 validation images, and 50 test images.
Languages
The class labels in the dataset are in English.
Dataset Structure
Data Instances
{
'image': <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=64x64 at 0x1A800E8E190,
'label': 15
}… See the full description on the dataset page: https://huggingface.co/datasets/Kanzoet97/Tinul.