FaceForm Studio — Public Portfolio Showcase

AI-powered personal style studio for face-shape analysis, virtual eyewear fitting, and hair-color preview.

About the project

FaceForm Studio is a portfolio project that combines machine learning, computer vision, backend development, and an interactive web interface.

The public showcase intentionally exposes only selected documentation, screenshots, configuration samples, and simplified code snippets. The production implementation, model weights, training notebooks, datasets, recommendation rules, and image-processing logic remain private.

What recruiters can see here

• Product UI and end-to-end user flow
• Face-shape analysis concept
• Virtual eyewear try-on
• Hair-color preview
• Technology stack
• High-level architecture
• Engineering practices such as testing, formatting, CI, and deployment planning
• Small sanitized code snippets that demonstrate coding style without exposing the implementation

Main features

1. Face-shape classification

   • EfficientNet-B0 based image classification
   • Five classes: Heart, Oblong, Oval, Round, Square

2. Virtual eyewear

   • Personalized frame recommendations
   • Face-landmark-based placement for visual preview

3. Hair-color preview

   • Hair-region processing
   • Multiple color options while preserving the existing hairstyle and texture

4. Product experience

   • Indonesian / English interface
   • Before / after comparison
   • Exportable preview result
   • Privacy-aware image handling

Selected screenshots

Face-shape analysis

Virtual eyewear

Hair-color preview

Technology stack

Area	Technology
Machine Learning	Python, PyTorch, TorchVision, EfficientNet-B0
Computer Vision	MediaPipe, Pillow
Backend	Flask
Frontend	Jinja2, HTML, CSS, JavaScript
Quality	Pytest, Black, isort, Flake8
Deployment	Docker, Gunicorn, Nginx, Redis

Repository scope

This public package is not the complete source repository.

Not included:

• app.py production implementation
• src/ production source
• model weights (.pth, .pt, .onnx, etc.)
• training notebooks
• datasets
• preprocessing and augmentation pipelines
• detailed recommendation mapping
• hair segmentation / recoloring implementation
• virtual try-on implementation
• private assets
• .env, API keys, credentials, deployment secrets

For the architecture summary, see docs/architecture-overview.md.

For the public/private boundary, see docs/public-private-boundary.md.

Demo

▶ Watch Full FaceForm Studio Demo

The demo shows the complete product flow, including face-shape analysis, virtual eyewear fitting, and hair-color preview.

The demo video is hosted externally on Google Drive so the public GitHub repository stays lightweight and focused on portfolio review.

Status

Portfolio showcase / active development

Author

Akhmad Syaifudin
Informatics Engineering — Universitas Islam Sultan Agung (UNISSULA)

Usage notice

This repository is intended for portfolio viewing and technical evaluation only.
No production source code, trained model, dataset, or internal implementation is distributed here.

See COPYRIGHT.md.