Skip to content

Datasets

Hugging Face datasets with plain-language descriptions: what's inside and what models to train on them. 4,788 in the catalog so far, with new ones added every hour.

Data type

For which models

Find a dataset for your goal

Answered by Gemini 4 · the chat is kept only in this tab

DatasetFreedomIntelligence/PubMedVision-PlusTop pick

Large medical image and caption corpus

Large-scale English dataset of tens of millions of medical image-caption pairs from PubMed articles.

  • Image → text
  • English
689 210M–100M rows
Datasethyperbots-inc/synthetic-doc-parsing

Synthetic annotated document parsing dataset

Dataset of 100 pages with fine-grained layout annotations, HTML tables, and text.

  • Image → text
  • Object detection
  • English
662 23up to 1K rows
Datasetgoogle/polaris-benchTop pick

Visual reasoning benchmark in polar coordinates

Official benchmark from Google DeepMind and Stanford for evaluating MLLM visual reasoning in polar coordinates.

  • Visual question answering
  • Multiple choice
  • English
627 710K–100K rows
Datasethyperbots-inc/synthetic-handwritten-forms

Synthetic handwritten-filled document forms

A dataset of synthetic printed forms filled in by hand for text recognition tasks.

  • Image → text
  • Document question answering
  • English
624 32up to 1K rows
Datasethyperbots-inc/synthetic-stamped-documents

Synthetic documents with stamps

A set of 100 synthetic business documents with colored stamps and seals for computer vision tasks.

  • Object detection
  • Image → text
  • English
593 30up to 1K rows
DatasetLitap007/indic-citizen-information-board

Multilingual Indian info board dataset

Multimodal text data in Indian languages for infrastructure auditing using VLM.

  • Visual question answering
  • Image → text
  • Table question answering
  • Hindi
  • Marathi
  • Telugu
  • +3
586 31K–10K rows
Datasetpriyank-m/chinese_text_recognition

Chinese text recognition dataset for images

A set of images with Chinese text (100K–1M examples) from the FudanVI benchmark for training OCR systems.

  • Image → text
  • Chinese
573 37100K–1M rows
Datasethyperbots-inc/synthetic-bad-scans

Synthetic degraded document dataset

120 corporate document images with various degradation levels for OCR testing.

  • Image → text
  • English
572 50up to 1K rows
DatasetCaptionEmporium/flickr-megalith-10m-internvl2-multi-captionTop pick

57.3M multimodal descriptions for the Megalith-10m base

A massive corpus of 57.3M short and long English captions for 10M Flickr images from various VLM models.

  • Text → image
  • Image → text
  • Other
  • English
548 331M–10M rows
DatasetKBlueLeaf/danbooru2023-metadata-database

Danbooru2023 Metadata in SQLite

An SQLite database of Danbooru2023 tags and posts with a latest ID of 7,866,491 for generative models.

  • Image classification
  • Text → image
  • Image → text
  • English
  • Japanese
522 841M–10M rows🔒 on request
Dataset5CD-AI/VietHTR-Line

60k lines of handwritten Vietnamese text for OCR

Manually verified image and transcription pairs of handwritten Vietnamese lines for OCR tasks.

  • Image → text
  • Vietnamese
510 2510K–100K rows🔒 on request
Datasetpriyank-m/SROIE_2019_text_recognition

Receipt text crops for OCR recognition

Over 52k cropped text fragments from English scanned receipts for training OCR models.

  • Image → text
  • English
504 1510K–100K rows
Datasethyperbots-inc/synthetic-rotated-text

Synthetic rotated documents dataset

Documents with text at various angles for OCR and recognition testing.

  • Image → text
  • Object detection
  • English
492 18up to 1K rows
Datasetfhswf/german_handwriting

German handwriting samples for OCR

About 10k images of handwritten German text with transcripts for training recognition systems.

  • Image → text
  • German
482 1910K–100K rows
DatasetCausalLM/Retrievatar

Multimodal Character RAG

100K synthetic image-text pairs about anime characters and celebrities for training RAG-equipped VLMs.

  • Image text → text
  • Image → text
  • Text → image
  • English
  • Chinese
  • Japanese
  • +1
419 75100K–1M rows
Datasetzhongshsh/CLoT-Oogiri-GO

Oogiri humor and meme multimodal dataset

Over 130,000 examples of Oogiri humor games with images and user ratings in three languages.

  • Visual question answering
  • Question answering
  • English
  • Chinese
  • Japanese
372 32100K–1M rows
Datasethyperbots-inc/synthetic-dense-documents

Synthetic dense document dataset

60 high-resolution pages with small text to test OCR robustness against complex structures.

  • Image → text
  • English
365 35up to 1K rows
DatasetHeffernan-WPI-Lab/DrawEduMath

Handwritten school mathematics solutions

Images of student handwritten math answers with teacher descriptions and QA pairs.

  • Visual question answering
  • English
348 11
DatasetVi-VLM/Vista

700k examples for Vietnamese VLMs

A large open multimodal dataset for training Vietnamese visual question-answering models.

  • Visual question answering
  • Vietnamese
345 45100K–1M rows
Datasetpathnirvana/sinhala-pali-ocr

Sinhala and Pali OCR Book Dataset

Scanned old Buddhist texts in Sinhala and Pali paired with proofread transcriptions for OCR.

  • Image → text
  • Sinhala
  • Pali
326 51K–10K rows
Datasetguangyil/laion-coco-aesthetic

Filtered LAION-COCO Aesthetic Subset

8.56M filtered English text-image pairs from LAION-COCO with aesthetic and watermark scores.

  • Image → text
  • Text → image
  • English
306 431M–10M rows
Datasetjainr3/diffusiondb-pixelart

2M Stable Diffusion generations with prompts

Large Stable Diffusion image corpus with user prompts and generation parameters.

  • Text → image
  • Image → text
  • English
297 611M–10M rows
Datasetalfredplpl/artbench-pd-256x256

Public domain art images

Filtered ArtBench subset with paintings at 256x256 resolution for image generation and captioning.

  • Text → image
  • Image → text
  • English
289 1010K–100K rows