Holo360D: 3D panoramic reconstruction with trajectories
Large-scale real-world dataset with continuous trajectories for panoramic 3D reconstruction.
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
Large-scale real-world dataset with continuous trajectories for panoramic 3D reconstruction.
Respiratory-synchronized CT scan dataset of lung cancer patients for segmentation.
A small dataset of bird song recordings from Doñana National Park for species detection and classification.
Dataset contains ResearchClawBench experiment artifacts: scientific reports, charts, code, and English metadata.
Compressed web models (.glb) of 1,033 real objects from Google Scanned Objects for 3D tasks.
English image and QA benchmark for testing AI agent memory over long sessions.
Popular scene graph generation benchmark with 150 object classes and 50 relation types in standard COCO format.
14,935 unique CAD tasks with images and three expert answers each.
300 sessions featuring first- and third-person video, speech audio, and text action labels.
An open portion of NVIDIA's Human-1M collection: first-person videos, head and hand poses with text descriptions.
Complex Olympiad-level problems with images in English and Chinese for evaluating model scientific reasoning.
Collection of aligned high-resolution portraits (NVIDIA's original includes 70k PNG 1024x1024).
Multimodal corpus (10K–100K) of posts and images regarding seven major natural disasters in 2017.
Processed DROID robotics dataset with optical and 3D scene flow annotations for training robot control models.
Unofficial mirror of the RainDrop (CVPR 2018) dataset for training models to clean lens water drops.
A prompt set to check the balance between LLM safety and their tendency to refuse harmless prompts.
Mirror of the well-known DTD dataset containing real-world texture photographs and annotations of their visual properties.
Design patent prior art search benchmark based on USPTO data.
Metadata and scripts for downloading bird images for visual question-answering tasks.
Multimodal dataset of 159,549 question-answer pairs for benchmarking in gastrointestinal endoscopy.
Multimodal evaluation dataset and MME benchmark from the MixLoRA paper (ACL 2024).
10 to 100 million arXiv text and image pairs for training research assistants.
A 750-sample English dataset subset configured for FiftyOne.
Document image set with physical distortions to test OCR and parsing system robustness.