Masked Face Age and Gender Identification

Image dataset containing artificially masked faces, annotated by age and gender, used for privacy-respecting facial recognition.

Download dataset

Size

Approximately 12,160 JPG/PNG images with artificial masks, total ~24,300 files

Licence

Apache 2.0

Description

‍

This dataset contains more than 12,000 images of human faces artificially masked by face masks, covering various age (0-116) and gender classes. It is the result of an augmentation method based on an original data set of 20,000 images. These images are annotated for age and gender identification in a privacy context, and are adapted for training deep learning models in computer vision.

‍

What is this dataset for?

‍

Training age and gender identification models with face masks
Develop facial recognition systems that respect confidentiality
Testing the robustness of computer vision algorithms on masked images

‍

Can it be enriched or improved?

‍

This dataset can be supplemented by finer annotations (expressions, ethnicities), or enriched by other types of increases. It is possible to create additional artificial masks with various styles to improve diversity. Manual correction of annotations and the addition of contextual metadata are also possible.

‍

🔎 In summary

Criterion	Evaluation
🧩 Ease of use	⭐⭐⭐✩✩ (Standard image format, requires preprocessing for some models)
🧼 Need for cleaning	⭐⭐⭐⭐✩ (Low to moderate: check annotation consistency)
🏷️ Annotation richness	⭐⭐⭐✩✩ (Annotates age and gender, no additional detailed annotations)
📜 Commercial license	✅ Yes (Apache 2.0)
👨‍💻 Beginner friendly	🌟 Yes, accessible for initial vision projects
🔁 Fine-tuning ready	🎯 Perfect for training and evaluating facial models
🌍 Cultural diversity	⚠️ Wide age range, moderate diversity depending on image source