Home robotics safety benchmark
1,000 simulations to verify how accurately and safely robots perform household tasks.
- Robotics
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
1,000 simulations to verify how accurately and safely robots perform household tasks.
Over 1.7 million source photo and instruction to target photo pairs for multimodal reasoning retrieval training.
About 25K YouTube frames with joint annotations for over 40K people for pose and action estimation.
Converted trajectory subsets from the OpenFly dataset for robotics applications.
A collection of metadata and image URLs with text descriptions for building and filtering training datasets.
Grid-aligned ocean data (temperature, salinity, surface height) across 50 depth levels for ML models.
Large corpus of screenshots and reworked interface annotations from web videos for GUI agents.
Over 20K document images across six languages for forgery detection and fake text analysis.
Multimodal logs of a physical dual-arm YAM robot with success and failure annotations.
Satellite data for methane emission segmentation, prepared for lightweight onboard neural networks.
Combined image dataset for training single-image deraining models.
Large-scale English corpus of image-text pairs extracted from open-access PubMed Central medical articles.
Millions of short daily first-person videos with detailed captions and Q&A from Reka AI.
Multiple-choice visual question answering data for two-stage VLM post-training.
Lightweight parquet files with links to ECMWF IFS weather forecasts on AWS S3 for analysis without downloading gigabytes.
2,595 web-ready .glb models from nvidia/simready-assets for three.js, Babylon.js, and other glTF 2.0 viewers.
Large-scale architecture photo base with metadata to train multimodal models in guessing the construction year.
Tens of millions of high-resolution Flickr images with English captions for image generation.
Media data and code agent run logs for the Embodied First Exam robot simulation benchmark.
English prompt dataset with image pairs for evaluating text-to-image preferences.
96 single-camera driving videos and over 105 thousand frames with depth and semantic segmentation annotations.
Multidisciplinary visual and textual questions for multimodal understanding evaluation.
9,209 everyday items, mechanisms, and CAD parts provided as sand3d code, OpenUSD (.usda) files, and renders.
English image and data collection from the VideoGrain paper (ICLR 2025) for multi-level video editing tasks.