Face-Depth-3D: RGB Images with Dense Depth Maps and 3D Meshes

Face-Depth-3D combines RGB face portraits with dense facial depth maps and 3D meshes in GLB format. The dataset, created by prithivMLmods, is designed for training and evaluating computer vision models. It was last updated on July 10, 2026.

Use Cases

• Train monocular depth estimation models based on the paired RGB images and depth maps.
• Supervise 3D face reconstruction networks using the provided 3D mesh ground truth.
• Evaluate the performance of image-to-3D models on a diverse collection of male and female face portraits.
• Benchmark algorithms for dense facial geometry prediction.

Strengths

• Each sample contains three aligned data modalities: an RGB image, a depth map, and a 3D mesh.
• The dataset includes both male and female face portraits, providing diversity in facial appearance.
• Data is provided in the GLB format, which is a common standard for 3D assets.

Limitations

• The total number of samples (rows) is unknown, which may limit suitability assessment for large-scale training.
• Column-level documentation is absent; field semantics must be inferred after download.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

huggingface

Freshness

Last updated 2026-07-10 04:18:55; freshness should be verified.