MultiShade: Synthetic Objects with Images and Normal Maps for 3D Rendering

MultiShade is a dataset for training, validating, and testing the RoSE model, as referenced for ICLR 2026. It contains images and corresponding ground truth normal maps for objects. The dataset was uploaded by author 'dennis98' and last updated on June 21, 2026.

Use Cases

• Training 3D reconstruction models based on provided object images and normal maps.
• Validating surface normal estimation algorithms using the ground truth normal maps.
• Testing the performance of the RoSE model on a dedicated benchmark dataset.

Strengths

• Includes paired data modalities: RGBA images and corresponding ground truth normal maps.
• Dataset is associated with a peer-reviewed conference submission (ICLR 2026).

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The dataset's creation method and source of objects are not described.

Provenance

Source

huggingface

Freshness

Last updated 2026-06-21 06:31:43; freshness should be verified.