datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
zendo-synthetic-data
Zendo Synthetic Visual Reasoning Dataset
Synthetic Zendo-style scenes with associated rules and per-scene tensor
representations. Each scene either follows ("positive", label=1) or violates
("negative", label=0) a rule that is given in natural language and as a Prolog
query.
Splits
split
scenes
train
56475
test
3344
rules total
3439
Layout
images/<split>/<batch>/<rule_id>/<scene_id>.png — rendered scene… See the full description on the dataset page: https://huggingface.co/datasets/sophia1ch/zendo-synthetic-data.synthetic-infrared-maritime-vessel-dataset-flux2klein
Synthetic Infrared Maritime Vessel Dataset (FLUX2-Klein)
RGB maritime vessel images translated to synthetic infrared via a DreamBooth-finetuned FLUX.2-Klein-4B.
Layout
synthetic-infrared-maritime-vessel-dataset-flux2klein/
├── in-distribution/
│ ├── train/{C00,C02,...}/*.jpg
│ ├── val/{C00,C02,...}/*.jpg
│ ├── test/{C00,C02,...}/*.jpg
│ ├── labels.txt
│ └── selected-metadata-{train,val,test}.json
└── out-of-distribution/
├── val/{C01,C03,C06… See the full description on the dataset page: https://huggingface.co/datasets/hanchong/synthetic-infrared-maritime-vessel-dataset-flux2klein.fashion_mnist
Dataset Card for FashionMNIST
Dataset Summary
Fashion-MNIST is a dataset of Zalando's article images—consisting of a training set of 60,000 examples and a test set of 10,000 examples. Each example is a 28x28 grayscale image, associated with a label from 10 classes. We intend Fashion-MNIST to serve as a direct drop-in replacement for the original MNIST dataset for benchmarking machine learning algorithms. It shares the same image size and structure of training and testing… See the full description on the dataset page: https://huggingface.co/datasets/zalando-datasets/fashion_mnist.NIH-Chest-X-ray-datasetThe NIH Chest X-ray dataset consists of 100,000 de-identified images of chest x-rays. The images are in PNG format.
The data is provided by the NIH Clinical Center and is available through the NIH download site: https://nihcc.app.box.com/v/ChestXray-NIHCCefficientnet-v2-l-adv-dataset
Perturb Adversarial Images
Verified adversarial examples for efficientnet_v2_l (torchvision/EfficientNet_V2_L_Weights.IMAGENET1K_V1), produced by the
Perturb network. Each row is one clean image together with all of its
verified adversarial versions: images that are imperceptibly different from the original
(L∞ ≤ 0.03 in [0,1] pixel scale) yet change the model's top-1 prediction.
This dataset grows continuously. New rows are appended as the network produces them and uploaded in… See the full description on the dataset page: https://huggingface.co/datasets/perturb-ai/efficientnet-v2-l-adv-dataset.MPII_Human_Pose_Dataset
Dataset Card for MPII Human Pose
MPII Human Pose dataset is a state of the art benchmark for evaluation of articulated human pose estimation.
The dataset includes around 25K images containing over 40K people with annotated body joints.
The images were systematically collected using an established taxonomy of every day human activities.
Overall the dataset covers 410 human activities and each image is provided with an activity label.
Each image was extracted from a YouTube… See the full description on the dataset page: https://huggingface.co/datasets/Voxel51/MPII_Human_Pose_Dataset.real-infrared-maritime-vessel-dataset
Real Infrared Maritime Vessel Dataset
Real infrared imagery of maritime vessels.
The dataset is provided in three forms — full-frame detection images, per-object classification crops, and a hand-curated subset.
Classes (7): liner, bulk carrier, warship, sailboat, canoe, container ship, fishing boat.
Layout
real-infrared-maritime-vessel-dataset/
├── original/ Full-frame IR images + XML bounding-box labels (detection)
│ ├── images/{train,test}/*.jpg… See the full description on the dataset page: https://huggingface.co/datasets/hanchong/real-infrared-maritime-vessel-dataset.military-aircraft-detection-dataset
Military Aircraft Detection Dataset
Military aircraft detection dataset in COCO and YOLO format.
The dataset was initially developed exclusively for military aircraft detection, but was later expanded to include commercial airliners for a broader and more challenging detection task.
The dataset contains 103 military aircraft types and 11 commercial airliner types.
Military aircraft: A10, A400M, AG600, AH64, AKINCI, AV8B, An124, An22, An225, An72, B1, B2, B21, B52, Be200, C1… See the full description on the dataset page: https://huggingface.co/datasets/a2015003713/military-aircraft-detection-dataset.autotrain-data-pick_a_card
AutoTrain Dataset for project: pick_a_card
Dataset Description
This dataset has been automatically processed by AutoTrain for project pick_a_card.
Languages
The BCP-47 code for the dataset's language is unk.
Dataset Structure
Data Instances
A sample from this dataset looks as follows:
[
{
"image": "<224x224 RGB PIL image>",
"target": 0
},
{
"image": "<224x224 RGB PIL image>",
"target": 0
}]
Dataset Fields… See the full description on the dataset page: https://huggingface.co/datasets/rwcuffney/autotrain-data-pick_a_card.MLLM-Generated-Image-Detection-Dataset
MLLM-Generated Image Dataset
This dataset contains real and AI-generated image samples organized for binary MLLM-generated image detection.
Paper | Code
Dataset Summary
We construct an MLLM-generated image detection benchmark from GPT Image2 and Nano Banana2. This benchmark covers texture-dominated, structure-dominated, and hybrid-dominated. It is designed to evaluate detector performance under the new challenges introduced by large-scale image generation models.… See the full description on the dataset page: https://huggingface.co/datasets/zr-zhang/MLLM-Generated-Image-Detection-Dataset.Describable-Textures-Dataset
Dataset Card for Describable Textures Dataset
This is a FiftyOne dataset with 5640 samples.
Installation
If you haven't already, install FiftyOne:
pip install -U fiftyone
Usage
import fiftyone as fo
import fiftyone.utils.huggingface as fouh
# Load the dataset
# Note: other available arguments include 'max_samples', etc
dataset = fouh.load_from_hub("Voxel51/Describable-Textures-Dataset")
# Launch the App
session = fo.launch_app(dataset)… See the full description on the dataset page: https://huggingface.co/datasets/Voxel51/Describable-Textures-Dataset.TALKtoME
TALKtoME: Educational Materials for Speech and Language Acquisition in Autism
This dataset contains image and video samples of action verbs and verb+noun pairs. It is designed to support machine learning tasks related to visual understanding, action recognition, and language grounding.
Dataset Structure
The dataset contains the following main folders:
action_verbs/images/: image samples organized by action verb categories.
action_verbs/videos/: video samples organized by… See the full description on the dataset page: https://huggingface.co/datasets/LSL-datasets/TALKtoME.Dataset
MM-OphBench: Multi-Center Multimodal Clinical Ophthalmic Benchmark Dataset
A Large-Scale, Standardized Multi-Center Benchmark Covering 7 Imaging Modalities & 4.3M+ Clinical Records
1. Executive Summary & Repository Overview
The MM-OphBench repository hosts a petabyte-scale, clinically harmonized ophthalmic image archive compiled from leading ophthalmic hospitals and benchmark cohorts. It spans 4,307,415 high-resolution diagnostic images and multimodal… See the full description on the dataset page: https://huggingface.co/datasets/Kaphathy/Dataset.Defactify_Image_Dataset
Defactify_Image_Dataset
This dataset is associated with the paper A Comprehensive Dataset for Human vs. AI Generated Image Detection.
📝 Dataset Description
Dataset Summary
The Defactify_Image_Dataset (A Comprehensive Dataset for Human vs. AI Generated Image Detection) is a high-quality collection of 96,000 images and associated metadata designed to benchmark models for detecting and identifying the source of artificially generated content. Built using the MS… See the full description on the dataset page: https://huggingface.co/datasets/Rajarshi-Roy-research/Defactify_Image_Dataset.UTM_Dataset
Dataset Description
This dataset was created to support Article: VisionGauge: a computer vision model to detect and read U-tube manometers and is available on GitHub
It consists of images of U-tube manometers constructed using a transparent PVC water level hose (5/16" × 1 mm) and flexible measuring tapes of different colors, each with a length of 150 cm (60 inches). The manometric fluids represented in the dataset include water, oil, and dyed water. The dataset is intended… See the full description on the dataset page: https://huggingface.co/datasets/claytonsds/UTM_Dataset.sph_dataset
SPH-Simulated LPBF Melt-Pool Dataset
Single-track laser powder bed fusion (LPBF) melt-pool simulations for Ti-6Al-4V,
produced with the LAMAS smoothed-particle-hydrodynamics solver. 241 simulations
sampled uniformly i.i.d. over a 4D process-parameter cube (laser power, scan
speed, laser spot radius, substrate temperature), spanning conduction, transition,
and keyhole regimes.
Companion to the NeurIPS 2026 Evaluations & Datasets Track submission
A Simulation-Based Dataset for… See the full description on the dataset page: https://huggingface.co/datasets/ioandanielc/sph_dataset.DDR-dataset
DDR - Diabetic Retinopathy Detection Dataset
Image: Dataset Samples.
The DDR (Diabetic Retinopathy Detection) dataset is a large-scale collection of retinal fundus images designed for training and evaluating algorithms in diabetic retinopathy (DR) grading and lesion-level segmentation. It provides both image-level DR labels and pixel-level annotations of pathological features, making it suitable for… See the full description on the dataset page: https://huggingface.co/datasets/ctmedtech/DDR-dataset.Masoud-Nickparvar-Brain-Tumor-MRI-Dataset
Dataset Card for Brain Tumor MRI Dataset
A dataset of 7,200 human brain MRI images, labeled into four classes — glioma, meningioma, pituitary tumor, and no tumor — for training and evaluating brain tumor classification models. This is a direct upload of the Brain Tumor MRI Dataset originally published on Kaggle by Masoud Nickparvar.
Dataset Details
Dataset Description
This dataset combines MRI images from three source datasets — figshare, the… See the full description on the dataset page: https://huggingface.co/datasets/BJyotibrat/Masoud-Nickparvar-Brain-Tumor-MRI-Dataset.autotrain-data-galaxy_classification
AutoTrain Dataset for project: galaxy_classification
Dataset Description
This dataset has been automatically processed by AutoTrain for project galaxy_classification.
Languages
The BCP-47 code for the dataset's language is unk.
Dataset Structure
Data Instances
A sample from this dataset looks as follows:
[
{
"image": "<256x256 RGB PIL image>",
"target": 0
},
{
"image": "<256x256 RGB PIL image>",
"target": 0
}]… See the full description on the dataset page: https://huggingface.co/datasets/Xanadu00/autotrain-data-galaxy_classification.qualcomm-interactive-video-dataset
Dataset Card for Qualcomm Interactive Video Dataset
This is a FiftyOne dataset with 2900 samples.
Installation
If you haven't already, install FiftyOne:
pip install -U fiftyone
Usage
import fiftyone as fo
from fiftyone.utils.huggingface import load_from_hub
# Load the dataset
# Note: other available arguments include 'max_samples', etc
dataset = load_from_hub("Voxel51/qualcomm-interactive-video-dataset")
# Launch the App
session = fo.launch_app(dataset)… See the full description on the dataset page: https://huggingface.co/datasets/Voxel51/qualcomm-interactive-video-dataset.cctv-datasets
CCTV Datasets for helmet detection + ANPR
Training and evaluation data used by vivekvar/helmet-v5 and vivekvar/helmet-v4.
Source: Andhra Pradesh RTGS CCTV feeds (public road cameras). All crops and frames are from motorcycle traffic scenes.
Folders
Folder
Contents
Purpose
merged_v3/
YOLO-format dataset (data.yaml + train/valid/test)
Bike + rider detection training
clean_merged_data/
Cleaned / deduped crop set
Base training data for v4
extra_khadatkar/… See the full description on the dataset page: https://huggingface.co/datasets/vivekvar/cctv-datasets.scanned-images-dataset-for-ocr-and-vlm-finetuning
Dataset Card for scanned_images_dataset
This is a FiftyOne dataset containing 3,482 scanned document images across 10 diverse document categories. Designed for OCR training and Vision-Language Model (VLM) fine-tuning, this dataset features real-world scanned documents with varied layouts, scanning quality, and document types.
Installation
If you haven't already, install FiftyOne:
pip install -U fiftyone
Usage
import fiftyone as fo
from… See the full description on the dataset page: https://huggingface.co/datasets/Voxel51/scanned-images-dataset-for-ocr-and-vlm-finetuning.car-damage-dataset
Car Damage Images
A raw image collection for vehicle damage assessment. Unlabeled: these images
have no annotations yet and are intended as source material for labeling or
pre-training.
Structure
images/
001/ images_001.jpg images_002.jpg ...
002/ ...
... 683 folders
thumbnails/
001/ thumbnail_001.jpg thumbnail_002.jpg ...
... 683 folders
Branch
Files
Folders
Size… See the full description on the dataset page: https://huggingface.co/datasets/Naiscorp/car-damage-dataset.isu-challenge-dataset
Dataset Card for ISU Challenge Dataset
Dataset Summary
ISU Challenge Dataset is a multi-modal in-cabin automotive dataset with a controlled synthetic core and paired real-reference scenes.
The synthetic core contains 1,000 synchronized Blender-rendered samples with:
RGB render
depth (EXR and PNG)
instance segmentation
canny edge map
structured scenario labels
An additional 60 paired real-reference scenes occupy sample_00000 through sample_00059. Each paired… See the full description on the dataset page: https://huggingface.co/datasets/ISU-Test/isu-challenge-dataset.Capillary-Dataset
Capillary dataset
Paper: Capillary Dataset: A dataset of nail-fold capillaries captured by microscopy for diabetes detection
Github: https://github.com/urgonguyen/Capillarydataset.git
The dataset are structured as follows:
Capillary dataset
├── Classification
├── data_1x1_224
├── data_concat_1x9_224
├── data_concat_2x2_224
├── data_concat_3x3_224
├── data_concat_4x1_224
└── data_concat_4x4_224
├── Morphology_detection… See the full description on the dataset page: https://huggingface.co/datasets/MelanieCo/Capillary-Dataset.MDS-Bench-data
MDS-Bench Raw Data
Raw input data for MDS-Bench, a benchmark that tests whether
AI agents can organize heterogeneous medical imaging datasets into a unified image + JSON format.
It contains 100 public medical imaging datasets (CT, MRI, microscopy, X-ray, ultrasound, endoscopy, ...;
DICOM, NIfTI, TIFF, H5, MAT, ...), about 2 TB in total.
Layout
origin_dataset/<dataset_name>/data/... # 8 datasets
second_dataset/<dataset_name>/data/... # 34 datasets… See the full description on the dataset page: https://huggingface.co/datasets/sdu-chenxin/MDS-Bench-data.LADI-v2-dataset
Dataset Card for LADI-v2-dataset
Dataset Summary : v2
The LADI-v2 dataset is a set of aerial disaster images captured and labeled by the Civil Air Patrol (CAP). The images are geotagged (in their EXIF metadata). Each image has been labeled in triplicate by CAP volunteers trained in the FEMA damage assessment process for multi-label classification; where volunteers disagreed about the presence of a class, a majority vote was taken. The classes are:
bridges_any… See the full description on the dataset page: https://huggingface.co/datasets/MITLL/LADI-v2-dataset.hateful-memes-data
Hateful Memes (CS5242 submission mirror)
Mirror of the Facebook Hateful Memes Challenge dataset (Kiela et al., 2020)
used for reproducibility of our CS5242 (NUS) submission.
Contents
img/ — 10,000 PNG images of memes
train.jsonl (8,500), dev_seen.jsonl (500), dev_unseen.jsonl (540),
test_seen.jsonl (1,000), test_unseen.jsonl (2,000)
Provenance
This mirror merges two existing mirrors of the original Meta release:
Label files and most images from… See the full description on the dataset page: https://huggingface.co/datasets/cs5242-hateful-memes/hateful-memes-data.ElectraAI-Dataset-v4
ElectraAI Dataset v4
1M+ record ECE/VLSI/Analog/Embedded dataset for LLM and vision model fine-tuning.
Statistics
Stream
Records
Description
A: NGSpice Netlists
100,000
SPICE netlists + analyses
B: Labeled Circuit Images
100,000
PNG + JSON (schemdraw)
C: QA Pairs (ShareGPT)
700,000
Multi-turn ECE instruction
D: YOLO Annotations
100,000
PNG + YOLO TXT (20 classes)
Total
1,000,000
Repo layout (sharded directories)
Images and… See the full description on the dataset page: https://huggingface.co/datasets/Abhisheksvnit/ElectraAI-Dataset-v4.llbench-dataset
LL-Bench: Rethinking Low-Level Vision Evaluation in the Era of Large-Scale Generative Models via Human Preferences
Anonymous release prepared for NeurIPS 2026 review. Please do not redistribute.
LL-Bench is a large-scale, human-preference benchmark for evaluating low-level
vision restoration in the era of large generative models (LGMs). It compares
10 LGMs with 16 specilist and 5 all-in-one models across 16 low-level vision tasks, paired with dense human annotations:pairwise… See the full description on the dataset page: https://huggingface.co/datasets/anonymousllbench/llbench-dataset.