DermaStyle AI: Intelligent Beauty & Virtual Try-On

Your AI beauty stylist—personalized skin analysis, smart product recommendations, and virtual clothing try-ons for confident, budget-friendly shopping.

Comment

DermaStyle AI: Intelligent Beauty & Virtual Try-On

Inspiration

We were inspired by the potential of artificial intelligence to transform beauty and fashion shopping into a more personalized, accessible, and confidence-building experience. Traditional shopping often leaves customers overwhelmed by countless products, uncertain about which skincare solutions suit their needs, and unable to visualize how clothing will look before purchasing.

We envisioned an intelligent beauty and fashion assistant that combines AI-powered skin analysis, personalized product recommendations, and virtual clothing try-ons in one seamless experience. Our goal is to help shoppers make informed decisions based on their preferences, shopping goals, and budget.

What it does

DermaStyle AI is an AI-powered beauty and fashion shopping assistant that brings personalized skincare recommendations and virtual fashion experiences together.

The platform enables users to:

  • Discover personalized recommendations: Describe their shopping goals, preferences, occasion, and budget using natural language.
  • Analyze skin with AI: Submit a facial image, with consent, for AI-powered skin analysis through Perfect Corp's YouCam API.
  • Find suitable products: Receive ranked product recommendations based on available skin-analysis results, personal preferences, and catalog information.
  • Experience virtual try-ons: Preview garments through AI-powered virtual clothing try-on technology.
  • Build a shopping plan: Generate budget-conscious product bundles tailored to individual shopping goals.
  • Track shopping sessions: Maintain session state and event logs throughout the recommendation workflow.
  • Protect personal images: Benefit from image validation, EXIF metadata removal, and a backend architecture that avoids writing uploaded photos to disk.

By combining these capabilities, DermaStyle AI aims to make beauty and fashion shopping more personalized, practical, and engaging.

How we built it

We developed DermaStyle AI using a modern full-stack architecture that integrates AI services with a responsive web application.

Backend

  • Python: Core application logic and AI integration.
  • FastAPI: REST API endpoints and asynchronous job workflows.
  • Pydantic: Data validation and structured models.
  • Pytest: Automated backend testing.

Frontend

  • Next.js: Web application framework.
  • React: Interactive user interface.
  • TypeScript: Type-safe frontend development.

AI and integrations

  • Perfect Corp YouCam API: AI-powered skin analysis and virtual clothing try-on.
  • Natural-language processing logic: Extracts shopping goals and preferences.
  • Recommendation engine: Ranks catalog products using available evidence and shopper preferences.
  • Budget-aware planning logic: Organizes recommendations into practical shopping bundles.

Data and privacy

  • A JSON-based sample product catalog supports the current demonstration.
  • Image validation checks file types, sizes, and minimum dimensions.
  • EXIF metadata is removed before images are submitted to YouCam.
  • API credentials remain on the backend.
  • Consent is required before facial or body analysis.

We also implemented automated API tests covering session management, skin analysis workflows, product recommendations, virtual try-on, and shopping-plan generation. The project's documented verification run completed with 30 passing backend tests.

Challenges we ran into

One of our main challenges was coordinating several distinct workflows—shopper profiling, image submission, AI skin analysis, product ranking, and virtual try-on—within a consistent user experience.

We addressed this by organizing the backend into dedicated modules and exposing API endpoints for each major stage of the shopping journey.

Other challenges included:

  • Asynchronous AI processing: Skin analysis and virtual try-on involve external API tasks that may take time to complete. We implemented job-status endpoints to support progress tracking.
  • Image quality and privacy: Facial and clothing images require appropriate validation and handling. We incorporated upload validation, EXIF normalization, and consent checks.
  • Meaningful recommendations: Product suggestions must reflect available evidence rather than unsupported skincare claims. We designed the ranking process to use catalog information and shopper preferences.
  • Budget-aware shopping: Recommendations need to balance individual preferences with affordability. We incorporated budget-oriented planning and product bundling.
  • External API dependencies: Live AI functionality depends on valid credentials, appropriate images, and successful responses from the external service.
  • Demo versus production readiness: The current implementation uses sample catalog data and in-memory session state, highlighting the need for persistent storage and real retailer integrations in a production deployment.

Accomplishments that we're proud of

We are proud to have developed a working full-stack foundation that brings AI-powered beauty analysis and virtual fashion experiences into one shopping workflow.

Our key accomplishments include:

  • Integrating Perfect Corp's YouCam API into a structured backend workflow.
  • Building a personalized shopping journey that connects user goals with product recommendations.
  • Implementing a catalog-based recommendation engine and budget-conscious shopping plans.
  • Creating asynchronous job workflows for AI analysis and virtual try-on.
  • Incorporating privacy-conscious image handling and explicit consent requirements.
  • Developing a modular FastAPI backend and a Next.js frontend.
  • Implementing automated API tests, with the documented verification run achieving 30 passing tests.
  • Establishing a foundation that can be extended with real retailer catalogs, persistent storage, and additional AI capabilities.

Most importantly, we transformed an idea into a testable application that demonstrates how beauty personalization and virtual fashion experiences can work together.

What we learned

Building DermaStyle AI reinforced the importance of treating AI as part of a complete product experience rather than as an isolated feature.

We learned that successful AI applications require more than connecting to a model or external API. They also need reliable data flows, robust validation, clear consent mechanisms, thoughtful user experiences, and recommendations grounded in available evidence.

We also gained practical experience in:

  • Integrating external AI services into a full-stack application.
  • Designing asynchronous workflows for time-consuming AI operations.
  • Structuring APIs around multi-step user journeys.
  • Connecting shopper preferences with catalog-based recommendations.
  • Incorporating privacy considerations into image-processing pipelines.
  • Testing backend endpoints and handling external-service dependencies.
  • Identifying the architectural improvements required to move from a local demonstration toward production deployment.

Above all, we learned that personalization is most valuable when it helps users make informed decisions while maintaining transparency about what the technology can and cannot determine.

What's next for DermaStyle AI: Intelligent Beauty & Virtual Try-On

Our next step is to evolve DermaStyle AI from a working demonstration into a scalable, production-ready beauty and fashion platform.

Our roadmap includes:

  1. Real retailer integrations: Replace the sample JSON catalog with live product data, accurate prices, stock availability, and retailer links.
  2. Persistent storage: Introduce a production database and durable session management to support returning users and reliable workflows.
  3. Enhanced AI personalization: Improve recommendations by combining product attributes, shopper preferences, budget constraints, and available skin-analysis results.
  4. Expanded virtual try-on: Refine the garment-preview experience and support a broader range of clothing categories.
  5. Smarter shopping assistance: Explore conversational AI that helps users compare products, understand recommendations, and refine their shopping goals.
  6. Stronger privacy controls: Expand consent management, data-retention controls, and mechanisms for users to clear session-derived information.
  7. Performance and reliability: Improve error handling, monitoring, caching, and resilience when external AI services are unavailable.
  8. User testing and evaluation: Assess recommendation relevance, usability, accessibility, and virtual try-on quality with real users.
  9. Responsible AI: Communicate the limitations of AI skin analysis, avoid presenting cosmetic assessments as medical diagnoses, and ensure recommendations remain transparent and evidence-based.
  10. Production deployment: Establish secure deployment pipelines, monitoring, and scalable infrastructure to support a growing user base.

Our vision: To make beauty and fashion shopping more intelligent, personalized, and accessible—helping every shopper explore suitable products, visualize their style, and make purchasing decisions with greater confidence.

Built With

Share this project:

Updates

Submission history