AI-generated image detection test benchmark
A collection of real and photorealistic AI-generated images across five domains for deepfake detector evaluation.
- Image classification
- 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
A collection of real and photorealistic AI-generated images across five domains for deepfake detector evaluation.
ChartQA dataset of 10k-100k examples with image and text modalities; card details are sparse.
A set of 1,505 visual questions to test multimodal model image reasoning capabilities.
Medical retina images annotated with diabetic retinopathy stages and pathologies.
Collection of game 3D objects in GLB format with pixel textures based on the Pixel Art Atlas.
115 real-world version upgrade tasks for open-source projects across 5 programming languages.
Image and Chinese text dataset for evaluating MLLMs in multi-image attribute extraction.
Over 8,000 synthetic English invoice scans partially annotated for OCR and document parsing.
A set of hyperspectral images (HSI) for non-destructive estimation of sugar content in processed foods.
An audio-video benchmark with masks for evaluating referring audiovisual segmentation quality.
A random sample of 5,000 multimodal satellite tiles of the contiguous United States for remote sensing tasks.
Mirror of the well-known ExDark low-light photography dataset, reformatted for YOLO.
Cleaned subset of the famous HaGRID dataset with 19,200 photos for 24 hand gesture recognition and detection.
Between 100K and 1M images for water-based object tracking from a Pattern Recognition paper.
49K English and Chinese questions about cats and dogs with images, audio, video, and text.
A dataset for testing vision-language models (VLM) on evaluating plan and route quality.
Synchronized video, audio, photo, and sensor data for human action recognition.
Image and text sets for testing models' text recognition capabilities.
20,580 dog photographs in English, prepared for image classification in FiftyOne format.
A challenging CVPR 2026 test set for evaluating text-to-image generation from multiple references.
450+ non-trivial questions with single-answer targets for testing tool-using and search-equipped LLMs.
Medical image dataset with questions, answers, and semantic annotations in English and Chinese.
Massive English dataset for Document VQA used in Idefics3 fine-tuning.
Hydrodynamic simulation graphs of cerebral aneurysms for training graph transformers.