CANDOR multimodal conversation corpus
Multimodal dataset of natural unscripted video conversations of strangers recorded during video calls.
- 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
Multimodal dataset of natural unscripted video conversations of strangers recorded during video calls.
102 episodes of xArm manipulator operation in LeRobot v3.0 format for training basic object grasping skills.
Hundreds of thousands of videos with English captions from Shanghai AI Lab for video models.
Video, time series, and joint telemetry of the Unitree G1 humanoid robot with Brainco hands during vegetable picking.
Test suite for evaluating AI agents' ability to write code for executable 4D scenes in Blender.
Head-mounted camera and LiDAR scanner recordings of people working in kitchens and auto repair shops in Southeast Asia.
Manually annotated video, audio, and images for evaluating deepfake detectors in real-world conditions.
Data from Alibaba and Chinese universities for training and evaluating models processing simultaneous video and voice streams.
First-person video benchmark with dense hand action annotations and sub-second segments starting from 100 ms.
Generated videos across various topics and aspect ratios to test MiniMax H3 capabilities.
1,355 episodes of pick and place task execution for robot training in LeRobot format.
About 134 hours of real-world robot operation scenarios for unfolding and manipulating crumpled T-shirts.
70 episodes of synchronous operation of two xArm arms in LeRobot v2.0 format for training bimanual tasks.
Test suite for evaluating AI agents that turn scripts or photos into animated 4D scenes in Blender.
Video and joint telemetry of the Unitree G1 humanoid robot with Inspire hands performing a vegetable picking task.
Dataset of SO-101 robotic arm simulation demonstrations drawing images using the LeRobot project.
Telemetry and 30 fps video logs ranging from 10k to 100k steps for training a robotic arm in cube transfer.
Football analytics video dataset with restricted access requiring an NDA signature.
1 to 10 million before-and-after video pairs with English text instructions for 10 editing tasks.
Dataset with videos and text questions from NVIDIA for road anomaly analysis, accepted at NeurIPS 2026.
A cleaned multimodal corpus for joint speech and video generation of talking persons in Chinese and English.
10,000 real driving video clips totaling over 80 hours with trajectory annotations and English text descriptions.
A multimodal real-world robotics task dataset for anomaly prediction.
Robotics demonstrations featuring time series and video for the sock folding task.