Discover, Connect & Collaborate at TECHINNOVATION 2021
Facial Weight Progression and Regression with Deep Learning
We present a framework to automatically perform a progressive/regression transformation of facial weight of an input face image to look heavier or thinner using deep-leaning based technology. Our framework is proven to generate high-quality and realistic facial-weight transformations without requiring a large amount of labelled image data. We offer a web application to show sample outputs of our framework using a variety of public datasets. Our framework can be incorporated into a health intervention that aims to motivate people to make healthier food choices by showing the potential consequences of their behaviour on appearance.
Technology Features, Specifications and Advantages
The proposed technology consists of a collection of facial image transformation solutions and web application, including but not limited to:
- Image processing module, including preprocessing and latent space embedding
- Facial weight transformation module
- Web application that allows the user to try out facial-weight transformations for a curated set of images.
The proposed technology is applicable to the following industrial scenarios:
Wellness & Healthcare
- Automated face transformation for visualizing the future impacts of their behaviour on appearance and motivating healthy food choices
- Intelligent food journal that visually translates energy intake balance to future weight gain/loss.
- Intelligent fitness tracking app that predicts facial weight change from physical activity data.
- Quick face shape editing tool and plugin for photo editing software
- The technology provides a live demonstration of an automatic transformation of a user’s portrait image to a thinner or heavier self
- The technology allows users to predict future facial look in food related application and lifestyle application based on food diary