Skip to content

Datasets

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.

Data type

For which models

Find a dataset for your goal

Answered by Gemini 4 · the chat is kept only in this tab

DatasetAPRIL-AIGC/Soul-Bench

Multimodal benchmark for avatar animation

Dataset of up to 1 thousand examples for evaluating image-to-video generation and digital human animation.

  • Image → video
730 56up to 1K rows
Datasetmarkov-ai/computer-use

OSWorld computer agent trajectories

160 successful Gemini agent trajectories on OSWorld tasks with steps and 1.0 scores.

  • Robotics
  • Image → text
728 77up to 1K rows
Datasetfpvlabs/karigar-100

Egocentric videos of skilled labor

Dataset of 102+ hours of egocentric workplace videos for robotics training.

  • Robotics
  • Video classification
728 21K–10K rows🔒 on request
Datasetsamliumay/FPV-Kamikaze-Drone-Action-Dataset

In-game FPV drone control dataset

FPV drone simulator gameplay recordings with timestamped controller states.

  • Robotics
719 21K–10K rows
DatasetHumynLabs/Street-videos

Street videos for detection and classification

Collection of street videos for training scene classification and object detection models.

  • Video classification
  • Object detection
  • English
621 470up to 1K rows
Datasetccmoony/PhyCoVE

Physically grounded video editing benchmark

508 tasks for altering physical object properties in video with reference clips and English/Chinese instructions.

  • Video → video
  • English
  • Chinese
615 2up to 1K rows
DatasetMichiganNLP/hmdb

HMDB51 video dataset

A dataset containing video files for human action classification tasks.

  • Video classification
  • English
598 310K–100K rows
DatasetInternRobotics/WET-OP-Bench-Train

Robot simulations for 14 scientific experiments

7,000 simulated Lift2 robot control episodes in LeRobot format for training text-instruction laboratory tasks.

  • English
593 21K–10K rows
DatasetMeituanRobotics/RoboDeskTop pick

Large video and episode dataset for robotics

Almost a million files and 6.49 TB of camera video recordings and episode tables for robot training.

    534 21K–10K rows
    Datasetego2act/ego2act-bench

    Ego2act benchmark for video generation

    A benchmark for evaluating goal-directed manipulations in video generation using egocentric frames.

    • Image → video
    • Video classification
    • English
    533 4up to 1K rows
    Datasetbinghong0727/egox

    EgoX: cross-embodiment manipulation

    Manipulation demonstration dataset from humans, real robots, and simulated humanoids in HDF5 and MP4 formats.

      524 31K–10K rows
      DatasetRapidata/sora-video-generation-physics-likert-scoring

      Physics Evaluation in Sora Video Generation

      198 generated videos with human physical realism scores for training and evaluating models.

      • Video classification
      • Text → video
      • English
      487 32up to 1K rows
      DatasetIntelLabs/Intel_Robotic_Welding_Multimodal_Dataset

      Multimodal robotic welding data

      Over 4k annotated welding samples: video, audio, sensor time series, and five seam images.

        423 99🔒 on request
        DatasetRapidata/sora-video-generation-style-likert-scoring

        Sora video visual style Likert scores

        198 generated videos with 6,021 human Likert scale ratings for visual appeal.

        • Video classification
        • Text → video
        • English
        382 34up to 1K rows
        DatasetHubdata/HubEgo-100KTop pick

        First-person video with manual work and IMU data

        Over 10,000 hours of first-person real-world workflows with motion sensors for training robots and multimodal models.

        • Robotics
        • Video classification
        • Video text → text
        • English
        335 510K–100K rows🔒 on request
        DatasetExploration-Lab/CISLR

        Indian Sign Language corpus (CISLR)

        A video dataset with text annotation for Indian Sign Language recognition from an EMNLP 2022 research paper.

        • sgn
        199 181K–10K rows🔒 on request