Human Face Images With Multiple Angles Light Conditions And Expressions

Featuring a sample of 110 people, with 2,100 images collected per subject. The images are structured across 14 camera angles, 5 light conditions, and 30 expressions.

Use Cases

• Train a facial expression recognition model using the 30 kinds of expressions as classification labels.
• Develop a face synthesis model conditioned on the 5 kinds of light conditions to generate images under varied illumination.
• Build a multi-view face reconstruction system leveraging the 14 kinds of camera angles for 3D modeling.
• Analyze the impact of different light conditions on the appearance of specific facial expressions.

Strengths

• Contains 2,100 images per subject, providing dense coverage for individual modeling.
• Systematically varies three key factors: 14 camera angles, 5 light conditions, and 30 expressions.
• Focuses on a specific demographic cohort of young people.

Limitations

• Limited to a sample of only 110 subjects, which may restrict model generalizability.
• The dataset is described as a sample of a paid dataset, implying the full data is not available.
• All subjects are young people, introducing a demographic bias.

Provenance

Source

Nexdata

Collection Method

Collected 2,100 images per subject, systematically varying camera angles, light conditions, and expressions.

Freshness

Last updated on 2025-04-24.

Geography

Region tag indicates 'us', but specific coverage is unknown.