datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
small
TempoFunk Small
7.8k samples of metadata and encoded latents & prompts of random videos.
Data format
Video frame latents
Numpy arrays
120 frames, 512x512 source size
Encoded shape (120, 4, 64, 64)
CLIP (openai) encoded prompts
Video description (as seen in metadata)
Encoded shape (77,768)
Video metadata as JSON (description, tags, categories, source URL, etc.)
CommunityForensics-Small
Community Forensics: Using Thousands of Generators to Train Fake Image Detectors (CVPR 2025)
Paper / Project Page / Code (GitHub)
This is a small version of the Community Forensics dataset. It contains roughly 11% of the generated images of the base dataset and is paired with real data with redistributable license. This dataset is intended for easier prototyping as you do not have to download the corresponding real datasets separately.
We distribute this dataset with a… See the full description on the dataset page: https://huggingface.co/datasets/OwensLab/CommunityForensics-Small.smart-bin-detect
arudaev/smart-bin-detect
Training data for Smart Bin Recognition – a validator ("is there a bin?")
and an identifier ("which bin?"). The design lives in docs/04-ml-pipeline.md
in the project repo, which is private; the manifests here carry per-image
provenance and are the authoritative record of what this dataset contains.
Every image carries provenance: source, source URL, licence, region,
capture date, annotator where known, label origin (human / machine /
legacy /… See the full description on the dataset page: https://huggingface.co/datasets/arudaev/smart-bin-detect.SmartHarvest
SmartHarvest: Multi-Species Fruit Ripeness Detection Dataset
Dataset Description
SmartHarvest is a comprehensive multi-species fruit ripeness detection and segmentation dataset designed for precision agriculture applications. The dataset contains high-resolution images of fruits in natural garden environments with detailed polygon-based instance segmentation annotations and ripeness classifications.
Key Features
8 fruit species: Apple, cherry, cucumber… See the full description on the dataset page: https://huggingface.co/datasets/TheCoffeeAddict/SmartHarvest.kaggle-chess-positions
Kaggle Chess Positions (mirror)
Public mirror of the Chess Positions dataset by koryakinp on Kaggle, hosted for easier access in ML pipelines (including 2d-chess-ocr).
Source & attribution
Field
Value
Original dataset
koryakinp/chess-positions
Original author
koryakinp
License
CC0-1.0
Generation tool
koryakinp/chess-generator
Board/piece assets
Chess.com
This repository is a mirror of the Kaggle release. Please cite the original Kaggle… See the full description on the dataset page: https://huggingface.co/datasets/smallchess/kaggle-chess-positions.dacl10k
Dataset Card for dacl10k
dacl10k stands for damage classification 10k images and is a multi-label semantic segmentation dataset for 19 classes (13 damages and 6 objects) present on bridges.
The dacl10k dataset includes images collected during concrete bridge inspections acquired from databases at authorities and engineering offices, thus, it represents real-world scenarios. Concrete bridges represent the most common building type, besides steel, steel composite, and wooden… See the full description on the dataset page: https://huggingface.co/datasets/smallopen1145141919810/dacl10k.Mobile_Phone_Dataset_Smartphone_and_Feature_Phone
Mobile Phone Dataset — Smartphone and Feature Phone (Sample)
⚠️ This is a free sample subset for evaluation purposes only.The full dataset (3,000+ HD images) is available for commercial licensing.Contact: sales@datacluster.ai · datacluster.ai
Dataset Summary
This dataset is an extremely challenging collection of original mobile phone images, crowdsourced from over 1,000 urban and rural areas. Every image is manually reviewed and verified by computer vision… See the full description on the dataset page: https://huggingface.co/datasets/Dataclusterlabspvtltd/Mobile_Phone_Dataset_Smartphone_and_Feature_Phone.MOUSS_fish_imagery_dataset_grayscale_small
Dataset Card for Modular Optical Underwater Survey System (MOUSS) Imagery - Small Set
This dataset contains grayscale underwater imagery collected by NOAA's Modular Optical Underwater Survey System (MOUSS), specifically for object detection of fish. The dataset is intended for training and evaluating models like the YOLOv8n-based Fish Detector on grayscale underwater footage.
Dataset Details
Dataset Description
This dataset is composed of black-and-white… See the full description on the dataset page: https://huggingface.co/datasets/akridge/MOUSS_fish_imagery_dataset_grayscale_small.Smart-Projector-Image-Classification-Dataset
Smart Projector Image Classification Dataset
With the rapid development of smart devices, a variety of portable projectors have emerged in the market. In practical applications, accurately recognizing and classifying the appearance images of these projectors has become a technical challenge. Current image recognition technology often yields poor classification results due to insufficient datasets or inaccurate annotations when dealing with different brands and models of projectors.… See the full description on the dataset page: https://huggingface.co/datasets/Mobiusi/Smart-Projector-Image-Classification-Dataset.MozzaVID_Small
MozzaVID dataset - Small split
A dataset of synchrotron X-ray tomography scans of mozzarella microstructure, aimed for volumetric model benchmarking and food structure analysis.
[Paper] [Project website]
This version is prepared in the WebDataset format, optimized for streaming. Check our GitHub for details on how to use it. To download raw data instead, visit: [LINK].
Dataset splits
This is a Small split of the dataset containing 591 volumes. We… See the full description on the dataset page: https://huggingface.co/datasets/dtudk/MozzaVID_Small.RealFakeDB_small#This dataset is collected from ImageReward for the fake class and COCO for the real class
fmow-fake-small
fmow-fake-small: a prototype dataset for remote sensing deepfake and image forgery localization
Dataset Details
Dataset Description
This dataset consists of 30 real and 30 manipulated satellite images. For the fake images, three types of image manipulations are used including random copy-paste splices, object copy-paste splices, and diffusion model inpainting. For the diffusion model inpainting, we use the pretrained diffusion model RSPaint.… See the full description on the dataset page: https://huggingface.co/datasets/geodf/fmow-fake-small.MV-VDB-photos-small
MV-VDB-photos-small
Media Vault - Vector Database Photos (Small)
A curated collection of 11,000 images from various computer vision datasets, designed for testing internal mechanisms in the Media Vault Vector Database system. This is the first small-scale dataset (targeting 10K samples, with NSFW split totaling 11K) for validation and testing purposes.
Dataset Structure
The dataset contains two splits:
sfw: All non-NSFW images (~10,000 images)
x_nsfw: Only NSFW images… See the full description on the dataset page: https://huggingface.co/datasets/SamoXXX/MV-VDB-photos-small.garbage-classification
Trash Classification - 5,000+ photos
Dataset comprises 5,000+ photos of garbage cans featuring various capacities, types, and waste materials, designed for advancing garbage classification and waste management systems. By leveraging this dataset, researchers and developers can enhance classification systems, automate garbage collection processes, and improve strategies for reducing environmental pollution. - Get the data
Dataset characteristics:
Characteristic… See the full description on the dataset page: https://huggingface.co/datasets/ud-smart-city/garbage-classification.Crowd-Countin-Dataset
Crowd Dataset - 647 Photos
Dataset comprises 647 photos of dense crowds, containing between 1,000 to 13,000 people per image. Each image includes detailed keypoint annotations for every individual, enabling advanced data analysis and deep learning applications in crowd density estimation, object detection, and counting algorithms. - Get the data
Dataset characteristics:
Characteristic
Data
Description
Crowd photos with labeling for determining crowd density… See the full description on the dataset page: https://huggingface.co/datasets/ud-smart-city/Crowd-Countin-Dataset.SAVANT-CODALM-small
SAVANT CODALM Small Dataset
This dataset is part of the SAVANT framework described in the SAVANT paper, currently under peer review.
This repository is provided for peer-review purposes only. After the review process, the dataset will be made publicly available through the authors' main account.
Dataset Description
CODALM small contains 100 real-world driving images (50 anomalous, 50 normal) derived from the CODA corner case dataset. Each image includes manual annotation… See the full description on the dataset page: https://huggingface.co/datasets/u94fmn391j/SAVANT-CODALM-small.Smart-Air-Conditioner-Image-Classification-Dataset
Smart Air Conditioner Image Classification Dataset
The current smart device industry faces challenges such as increased complexity in user interaction and insufficient accuracy in device recognition. Existing solutions often rely on traditional image recognition technology, resulting in low accuracy and poor adaptability. The construction of the Smart Air Conditioner Image Classification Dataset aims to address this issue by providing high-quality image data to help models better… See the full description on the dataset page: https://huggingface.co/datasets/Mobiusi/Smart-Air-Conditioner-Image-Classification-Dataset.index-card-blank-content
Index-card blank / content / divider classifier — dataset
Cropped single archival index cards labelled blank, content, or divider, for
training a tiny CPU pre-filter that skips blank/divider cards before expensive VLM metadata
extraction in card-catalogue digitisation pipelines.
Two collections: Boston Public Library (BPL) FRC shelf-list cards and National Library
of Scotland (NLS) Advocates Library cards. Styles differ, so evaluate per collection.
How it was made… See the full description on the dataset page: https://huggingface.co/datasets/small-models-for-glam/index-card-blank-content.Smart-Speaker-Image-Classification-Dataset
Smart Speaker Image Classification Dataset
The current smart speaker market is highly competitive with diverse brands and forms, leading to difficulties for consumers in identification when making purchases. Existing image datasets mostly focus on general products, lacking classification datasets specifically for smart speakers, resulting in obvious deficiencies in brand identification and appearance classification. This dataset aims to help develop more efficient smart speaker… See the full description on the dataset page: https://huggingface.co/datasets/Mobiusi/Smart-Speaker-Image-Classification-Dataset.smash-or-transformer-data
Smash or Transformer -- mixed_v1 dataset
The training dataset for the recommended
Smash or Transformer model:
21,076 images across all 1,025 Pokemon (official artwork + in-game sprites +
Safebooru fan-art), with crowd "smash" labels from pokesmash.xyz.
Code, docs, and reproduction guide:
https://github.com/byrte1024/SmashOrTransformer/blob/master/docs/vit_small_mixed_v1.md
Contents
mixed_v1_dataset.tar (~11 GB) extracts at the repo root to:
datasets/mixed_v1/ --… See the full description on the dataset page: https://huggingface.co/datasets/supernovayuli/smash-or-transformer-data.image-censorship-small
Image Censorship Dataset
Датасет разработан специально для корректного бенчмаркинга детекции NSFW-контена.
Изображения подобраны так, чтобы распределение было схоже с реальным user input | generated output.
Датасет представляет собой осмысленную компиляцию существующих с разметкой safe | unsafe.
Схема
поле
тип
значения
image
Image
RGB
label
ClassLabel
safe / unsafe
is_edge_case
bool
пограничные примеры (sexy, spa, rescue, voter)
category
string… See the full description on the dataset page: https://huggingface.co/datasets/vekshinkir/image-censorship-small.viewer-test-50-small-images-ok
Test Dataset: viewer-test-50-small-images-ok
Purpose: Demonstrate dataset viewer behavior with large images and row group sizing.
Dataset Details
Number of images: 50
Total size: ~167MB
Average per image: ~3.3MB
Expected Behavior
50 smaller images (~3.5MB each, ~175MB total). Well under the 300MB limit.
Prediction: SHOULD WORK - Under 300MB limit
Technical Context
The dataset viewer converts imagefolder datasets to parquet with:
Row group size:… See the full description on the dataset page: https://huggingface.co/datasets/davanstrien/viewer-test-50-small-images-ok.text-2-image-dpo-human-preferences-small
Text-2-Image DPO Human Preferences (Small)
A quality-controlled human preference dataset for text-to-image generation. 40,000 trust-weighted pairwise judgments from calibrated annotators comparing AI-generated images across two evaluation dimensions: prompt alignment and overall preference.
This is the highest-annotator-quality subset. For the full 5,000-pair dataset, see datapointai/text-2-image-dpo-human-preferences.
Built on the Datapoint annotation platform — purpose-built… See the full description on the dataset page: https://huggingface.co/datasets/datapointai/text-2-image-dpo-human-preferences-small.SmartVision-Dataset
SmartVision Classification Dataset
Classification dataset used by the SmartVision AI Platform.
Structure: