Face Shape Classification Dataset
Dataset of 5000 images of female faces, classified into 5 categories according to the shape of the face: heart, oval, round, oblong and square.
5000 JPG images divided into 5 classes of face shapes (heart, round, oval...)
CC0: Public Domain
Description
The Face Shape Classification Dataset includes 5000 images of female faces of celebrities from around the world, divided into 5 distinct facial shapes: heart, oblong, oval, round, and square. Each class contains 1000 images, divided between a training set (800) and a test set (200).
What is this dataset for?
- Develop facial shape recognition algorithms
- Create personalized recommendation systems for hairstyling or eyewear
- Training generative AI models or facial aesthetics
Can it be enriched or improved?
This dataset can be enriched by adding additional annotations such as ethnicity, age, or facial symmetry. Applying image augmentation techniques (rotation, lighting) can also improve the robustness of models.
🔎 In summary
🧠 Recommended for
- AI-assisted aesthetics projects
- Biometric search
- Facial classification
🔧 Compatible tools
- OpenCV
- PyTorch
- Keras
- Tensorflow
- FastAI
💡 Tip
Train a basic model with all 5 classes, then refine for personalized recommendations based on facial attributes.
Frequently Asked Questions
Are images labelled automatically or manually?
The categories are pre-assigned, probably manually, according to the shape of the face observed.
Can this dataset be used for a commercial beauty recommendation project?
Yes, the dataset is under a CC0 license, so it can be used without commercial restrictions.
Does the dataset cover a diversity of genres or origins?
It focuses on famous female faces; expanded diversity would require enrichment from other sources.
