Datasets:

Vigen1
/

hypersim_big

Tasks:

Image Segmentation

Modalities:

Formats:

Size:

Tags:

computer-vision

Libraries:

License:

Dataset card Data Studio Files Files and versions

hypersim_big / README.md

Vigen1's picture

Upload README.md with huggingface_hub

4876613 verified 6 months ago

|

history blame contribute delete

682 Bytes

datasets:
  - Vigen1/hypersim_big
tags:
  - computer-vision
  - segmentation
task_categories:
  - image-segmentation
pretty_name: Hypersim Minimal (RGB + Semantic Mapped)
license: other

Hypersim Minimal (RGB + Semantic Mapped)

This dataset is a minimal extraction from Hypersim:

RGB: preview (either preview JPGs or HDR color.hdf5 tonemapped to PNG)
Semantic labels: mapped to uint8 PNG using clip40to39

Columns

image — RGB image
mask — segmentation mask (uint8)
scene, camera, frame — identifiers

Splits

train
validation (ratio: 0.1)

Note: You are responsible for complying with the original Hypersim license/terms.