VLM spatial reasoning benchmark in mirror scenes
Over 14,000 question-answer pairs on real photos with reflections for evaluating multimodal models.
- Visual question answering
- English
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.
Answered by Gemini 4 · the chat is kept only in this tab
Over 14,000 question-answer pairs on real photos with reflections for evaluating multimodal models.
Large dataset of image-LaTeX pairs for mathematical formula recognition.
Port of the popular ImageNet100 dataset with images and text labels for computer vision tasks.
Between 100k and 1M SVG icon examples for training text-to-image and vector graphics generation models.
Over 1.5 million ratings from real people for images generated from English text prompts.
A large corpus of labeled images for benchmarking federated learning (FL) models.
Corpus of 100K to 1M poems by modern Chinese and foreign poets with a selection for image-based generation.
Labeled dataset of Persian license plate images and text for OCR tasks.
Detailed synthetic descriptions of 16.9 million images for training and evaluating vision-language models.
A compact image dataset from the MiniGPT-4 project designed to align visual data with language models.
A test suite for evaluating multimodal agents working with graphical user interfaces across six platforms.
A large dataset of English graph annotations for CC12M images tailored for multimodal models.
High-precision aerodynamic simulation dataset for training physics AI models in automotive engineering.
10,000 real and AI-generated images labeled across 11 categories of unsafe content.
Dataset of up to 1 thousand examples for evaluating image-to-video generation and digital human animation.
160 successful Gemini agent trajectories on OSWorld tasks with steps and 1.0 scores.
FPV drone simulator gameplay recordings with timestamped controller states.
Synthetic dataset with images generated by GPT-4o to improve text-to-image models.
Photos of assembled electrical panels with 14 types of manufacturing defects for anomaly detection.
A collection of images and masks for evaluating analog media damage segmentation models.
Large-scale English dataset of tens of millions of medical image-caption pairs from PubMed articles.
Images for training and evaluating NSFW detectors in artistic content; five classes.
Structured descriptions, ratings, tags, and video game metadata from the RAWG API for analytics and recommendation systems.
Dataset of 100 pages with fine-grained layout annotations, HTML tables, and text.