fIMDb

fIMDb indexes and annotates face image databases to enable selection of representative datasets for experimental stimuli and benchmarking of face recognition algorithms.

Key Features:

• Database index: A comprehensive index of known face databases with metadata detailing dataset features and access methods.
• Demographic filters: Filters to locate databases by race, ethnicity, and age group.
• Visible-difference metadata: Metadata indicating presence of visible differences such as scars, port wine stains, and cleft lip and palate.
• ChatLab Facial Anomaly Database (CFAD): Inclusion of CFAD with photographs of faces exhibiting various visible differences annotated across etiology, size, location, ethnic background, and age.
• Dataset selection support: Facilitation of selection of datasets for experimental stimuli creation and benchmarking of face recognition algorithms.

Scientific Applications:

• Experimental stimuli creation: Selection of representative and demographically diverse facial stimuli for behavioral and neural studies.
• Algorithm benchmarking: Benchmarking and evaluation of face recognition and computer vision methods across diverse demographics and visible differences.
• Perceiver response research: Studies of perceivers' attitudes, behaviors, and neural responses to faces with visible differences.
• Domain research support: Use in psychology, neuroscience, and computer vision research requiring well-characterized face image datasets.

Methodology:

Indexes known face databases and annotates metadata (dataset features, access methods, demographics, visible differences), implements filters for demographic and visible-difference attributes, and compiles CFAD photographs annotated for etiology, size, location, ethnic background, and age.