Aakash941/SAFFIRE-Face-Dataset

SAFFIRE (South Asian Facial Features in Real Environments) is a high-quality annotated facial dataset designed for age, gender, occlusion, and pose estimation. It comprises diverse South Asian faces captured in real-world conditions, ensuring variability in lighting, background, and expressions. The dataset includes detailed annotations for facial attributes, making it valuable for robust AI model training.

For further information and citation please refer to the paper: SAFFIRE: South Asian Facial Features in Real Environments- a high-quality annotated facial dataset for Age, Gender, Occlusion, and Pose estimation

There are 2 files in the dataset: 1st PK-face.zip and 2nd Face-annotation.csv. The first file is a zipped compression of a folder containing actual facial images. While the later contains the details of annotation recorded for those facial images. The CSV file contains 6 columns details are as follows: ImageID: Unique ID of the image Path: Physical path of the image inside the uploaded folder Gender: Assigned gender to the facial image (Male or Female) Age-group: Assigned age group to the facial image (Juvenile: 0-18, Adult: 19-59, Senior: 60+) Pose: Assigned pose of the facial image (Frontal, Three-Quarter, Profile) Occlusive: Signifies whether the facial image is clear or occlusive in nature (Yes or No)