Skip to content

Datasets

Hugging Face datasets with plain-language descriptions: what's inside and what models to train on them. 4,797 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

Datasetvivym/midjourney-prompts

Cleaned Midjourney Prompts Across Versions

Over 9 million cleaned text prompts for the Midjourney image generator versions 1.0 through 5.2.

  • Text → image
  • English
265 541M–10M rows
Datasetkonfuzio/funsd_plus

FUNSD+ dataset for form understanding

An expanded and corrected version of the FUNSD dataset for document information extraction.

  • English
245 51K–10K rows
Datasetsujet-ai/Sujet-Finance-QA-Vision-100k

Visual question answering for financial documents

Over 107,000 English question-answer pairs for nearly 10,000 images of various financial documents.

  • Question answering
  • English
226 401K–10K rows
Datasetvavaw/SkyLumeTop pick

SkyLume urban UAV dataset

Large-scale real-world urban UAV dataset for 3D reconstruction under varying illumination.

  • Image → 3d
225 23🔒 on request
DatasetOpenFace-CQUPT/FaceCaption-15M

15 million face and text caption pairs

A large English dataset of facial images with captions for description and generation.

  • Image → text
  • Text → image
  • English
223 10710M–100M rows
DatasetRapidata/bananamark-dataset

Comparison of 11 image generators on banana prompts

A dataset of human pairwise evaluations (2,161 rows) for banana images from 11 generative models.

  • Text → image
  • Image → text
  • Image classification
  • English
216 351K–10K rows
Datasetfacebook/emu_edit_test_set_generations

Emu Edit image editing generation examples

Generation results from Meta's Emu Edit model on a test set for comparison and quality evaluation.

    206 381K–10K rows
    Dataset5CD-AI/Viet-Handwriting-OCR-v2

    Vietnamese handwriting OCR

    Image-to-text data for recognizing Vietnamese handwriting; the card promises a fuller version.

    • Image → text
    • Vietnamese
    205 10510K–100K rows🔒 on request
    DatasetKORMo-VL/Ko-Screenspot

    Ko-Screenspot Korean GUI grounding benchmark

    A dataset of 3,319 screenshot-instruction pairs for GUI grounding tasks on Korean websites.

    • Image text → text
    • Korean
    199 64up to 1K rows
    Dataset5CD-AI/LLaVA-CoT-o1-Instruct

    LLaVA-CoT: visual reasoning instructions

    Image-question-answer pairs with step-by-step reasoning for training multimodal CoT models.

      198 11310K–100K rows🔒 on request
      Datasethuggan/pokemon

      Pokemon Image Dataset

      Collection of Pokemon images for computer vision and generation tasks.

        195 241K–10K rows
        DatasetOpenFace-CQUPT/HumanCaption-10MTop pick

        10 million human image captions

        Large-scale English image-text pair dataset for generating and recognizing images of people.

        • Image → text
        • Text → image
        • English
        190 4810M–100M rows
        Datasetyandex/mad-carsTop pick

        Yandex 360-degree car reviews (MAD-Cars)

        A large dataset of ~70K full HD circular car video reviews for video generation tasks.

        • Image → video
        188 371M–10M rows
        Datasetsoymia/boudoir-dataset

        Behance gallery images for image generation

        Selection of photographs from chosen Behance galleries with text data for text-to-image tasks.

        • Text → image
        186 161K–10K rows
        Datasetdclure/laion-aesthetics-12m-umap

        UMAP Coordinates for 12M LAION Aesthetic Images

        Tabular UMAP projection coordinates of text CLIP embeddings for the LAION-Aesthetics sample with scores above 6.

        • English
        184 4110M–100M rows
        DatasetJaviai/failures-3D-print

        3D printing defect images

        A small set of annotated photographs of 3D printing errors for object detection tasks.

        • Object detection
        181 6up to 1K rows