Video-to-Video
video-to-data-robot-dexterity-task-library-and-dataset
Video to Data: Robot Dexterity Task Library and Dataset
Dataset Description
This dataset contains samples of human demonstrations on manipulation tasks retargeted to bimanual Sharpa robot hands and episodes of robot executions that mimic the original human demonstrations. The former allows a Video to Data user to easily experiment with the Video to Data grounding pipeline, and the latter is an example of the grounded robot data that can be generated with the Video… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/video-to-data-robot-dexterity-task-library-and-dataset.video_to_data_challenge
Video to Data (V2D) Challenge Dataset
Dataset Description
The Video to Data (V2D) Challenge Dataset is an NVIDIA-developed benchmark for studying the complete path from human demonstration video to physics-grounded robot behavior. It supports three coupled challenge tracks over shared manipulation tasks: 4D human-object interaction reconstruction, robotic grounding, and end-to-end egocentric transfer.
Challenge website
Starter toolkit
Dataset repository
Contact:… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/video_to_data_challenge.video-to-data-object-assets
NVIDIA Video-to-Data Object Assets
NVIDIA Video-to-Data Object Assets is a collection of textured 3D object meshes and
simulation-oriented USD packages produced for the
NVIDIA Video-to-Data project.
The repository is hosted as a Hugging Face dataset for versioned distribution, but its contents
are 3D assets rather than raw recordings, an annotated machine-learning dataset, or a benchmark.
Version and contents
This is the V2D v0.3 object-asset release.… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/video-to-data-object-assets.video-to-data-robot-dexterity-task-library-and-dataset
Video to Data: Robot Dexterity Task Library and Dataset
Dataset Description
This dataset contains samples of human demonstrations on manipulation tasks retargeted to bimanual Sharpa robot hands and episodes of robot executions that mimic the original human demonstrations. The former allows a Video to Data user to easily experiment with the Video to Data grounding pipeline, and the latter is an example of the grounded robot data that can be generated with the Video… See the full description on the dataset page: https://huggingface.co/datasets/introvoyz042/video-to-data-robot-dexterity-task-library-and-dataset.image-to-video-human-preference-seedance-1-pro
Rapidata Video Generation Hailuo-02 v Marey Human Preference
In this dataset, ~6k human responses from ~2k human annotators were collected to evaluate Seedance 1 Pro video generation model on our benchmark. This dataset was collected in roughtly 5 min using the Rapidata Python API, accessible to anyone and ideal for large scale data annotation.
Explore our latest model rankings on our website.
If you get value from this dataset and would like to see more in the future, please… See the full description on the dataset page: https://huggingface.co/datasets/Rapidata/image-to-video-human-preference-seedance-1-pro.commercial-image-to-video-model-benchmark
Commercial Image-to-Video Model Benchmark via Magic Hour
This reproducible study evaluates commercially useful image-to-video workflows across leading models available through one Magic Hour API integration.
Permanent archived release: Zenodo record and DOI 10.5281/zenodo.22683989.
Magic Hour AI, Inc. designed and funded the study and provides the API used to access every evaluated model. Results are published with all attempts, observed terminal time, API credits charged… See the full description on the dataset page: https://huggingface.co/datasets/magichourhq/commercial-image-to-video-model-benchmark.