Multimodal PDF corpus for VLM pre-training
Part of the giant open MINT-1T dataset featuring interleaved English text and PDF images.
- Image → text
- Text generation
- English
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.
Answered by Gemini 4 · the chat is kept only in this tab
Part of the giant open MINT-1T dataset featuring interleaved English text and PDF images.
Tens of millions of text image and transcript pairs for training Telugu optical character recognition systems.
An image dataset for cell detection and classification tasks in microscopic images.
A collection of Chinese technical tables and blueprint images for evaluating Multimodal Large Language Models (MLLMs).
24,370 images with 470,000 annotated people for evaluating object detectors.
34,627 images of sign language letters adapted for use with the FiftyOne library.
A set of text and visual tasks to evaluate AI agent capabilities in scientific biology.
A multimodal dataset with before-and-after disaster images and 123k instructions for VLM models.
A collection of 15,620 interior photos annotated across 67 categories in FiftyOne format.
A set of 8,629 drone images for training and validating computer vision models.
A set of 100 3D models with English text prompts and annotations for evaluating localized 3D editing.
Image-text pairs for training multimodal models to detect image editing artifacts.
Dataset containing geometry descriptions, parameters, images, and FreeCAD files for CAD modeling.
25,435 triples of scientific papers, posters, and HTML/CSS templates from AI conferences for poster generation.
Visual benchmark for egocentric-exocentric action understanding and future state prediction.
10k to 100k synchronized stereo, RGB, IMU and tactile recordings for 11 manipulation tasks.
Part of a giant open dataset with interleaved texts and images from PDFs and the web.
Field images of electrical distribution infrastructure with annotations for detection, segmentation, and classification.
Over 100,000 urban scenes in OpenUSD format with textures for robot training.
A set of 49,706 annotated images with English metadata for image-classification and object-detection tasks.
Over 1.4 million screenshots with detailed screen element annotations for training UI parsing models.
English multimodal test suite for checking how well vision-language models reason in robotics scenarios.
99 hours of complete engineer workflows in desktop CAD software with video, text, source files, and evaluation criteria.
Image-text description pairs for training and evaluating vision-language models.