Occluded And Multi Pose Face Recognition Sample

Aggregating a sample of 1,930 people, with 200 images collected per subject. The images feature 4 lighting conditions, 10 occlusion cases, and 5 face poses. It is designed for computer vision tasks like occluded face detection.

Use Cases

• Train models for occluded face detection using the 10 distinct occlusion case labels.
• Develop multi-pose face recognition systems leveraging the 5 specified face pose variations.
• Analyze the impact of 4 different lighting conditions on face recognition model performance.
• Benchmark robustness of recognition algorithms against combined variations in occlusion, pose, and lighting.

Strengths

• Contains 1,930 distinct subjects, providing a variety of identities.
• Offers structured variation with 4 lighting conditions, 10 occlusion cases, and 5 poses per subject.
• Provides 200 images per subject, enabling within-subject analysis of multiple factors.

Limitations

• The dataset is explicitly a sample of a larger paid dataset, limiting its completeness.
• Specific image counts, file formats, and column metadata are not provided in the input.
• Potential geographic or demographic bias is unknown as region details are unspecified.

Provenance

Source

Nexdata

Collection Method

Collected images per subject under controlled variations of lighting, occlusion, and pose.

Freshness

Last updated on 2025-04-24.

Geography

Region tag suggests 'us', but specific spatial coverage is not detailed.