- Home
- Success Stories
- AI-ML Powered Face Morphing & Age Progression Application
Developed an ML-based face morphing and age progression application
Introduction
Helped a leading US biopharma company develop a machine learning (ML) powered face morphing solution, enabling patients to visualize wrinkle progression, seek timely care, and improve treatment outcomes. The region-of-interest detection algorithm achieved 90% accuracy.
Skincare is personal. For a leading US-based biopharmaceutical company focused on specialty medicines and treatments, empowering patients to make more informed decisions around skin health was both a mission and an opportunity.
The goal: Develop an intuitive, AI-powered experience that could analyze skin texture, detect early signs of aging, and simulate wrinkle progression, all tailored to individual users.
To bring this idea to life, the client turned to CitiusTech.
The Challenge: Accurate skin analysis across tones, types, and time
The company wanted to give users a simple tool to assess their skin’s wrinkle stage and explore treatment options with greater confidence. But achieving this required solving a complex technical problem: how to detect subtle discontinuities in skin texture, across diverse skin surfaces and tones, and project future wrinkle progression precisely.
They needed an algorithm that was not just technically sound but also free of bias and optimized for real-world use.
The Solution: AI-driven face morphing with fairness at its core
Developed a predictive algorithm rooted in dermatological insight and precision modeling. The team utilized public datasets with clustering-based ML algorithms and statistical concepts for wrinkle measurement, pre-trained models for facial landmark detection, and GAN-based models for age progression functionality.
Key capabilities included:
- A skin wrinkle detection engine built using prior knowledge and deterministic modeling.
- An iOS app was developed to capture a face image, show wrinkle analysis, and provide face-image timelines.
- A Python Flask framework was used to integrate the ML functionality into the application.
- HDBSCAN clustering-based algorithm, OpenCV filters were used for accurate wrinkle-scale measurement.
- A scale detection model was used to classify the glabellar region wrinkles into 4 classes based on severity.
- Random Forest classifier, along with Gabor features, was used to accomplish hyperparameter tuning, using the sklearn library.
- A user-friendly front-end to engage patients in their care journey.
The solution was designed to balance accuracy with user empowerment, enabling consumers to assess changes in their skin and explore medical treatment paths with confidence.
Fig 1: Solution Schematic
Value Delivered: Empowering users, enhancing engagement
With this face morphing solution, the biopharma client was able to deliver an innovative experience that put predictive care in the hands of users:
- Achieved 90% accuracy, proven over 60+ unique images. Evaluated using the IOU (Intersection over union) concept
- Enabled patients to self-analyze wrinkle stages and seek timely remote consultations
- Used predictive modeling to show future age-progression imagery and improve outcomes
- Removed prediction bias by evaluating models against industry-standard fairness metrics
- Provided an integrated solution for patient engagement that is focused on user empowerment and satisfaction
By translating AI into personalized insights, the solution helped the client go beyond treatment.
Healthcare AI that’s personalized, predictive, and powerful
This project reflects CitiusTech’s strength in building intelligent, human-centered AI solutions for healthcare. It delivered an integrated solution that redefines digital health engagement with a blend of domain expertise, technical depth, and patient-centric design.