BeautyDeep

BeautyDeep is an app which evaluates the beauty of a human face from a photo given. The system itself recognizes one or several faces on the photo and evaluates the percentage of their beauty, also analysing each persons character according to their face features.

The program uses a neural network trained to evaluate the beauty percentage on the basis of more than 10 thousand photos already rated by real people, and also detectors recognising the features of the face. Using mathematical formulas the face features are typified, each of the types describing the persons character according to neurotypology laws.

Project Preview

Project Pipeline

BeautyDeep project consists of an Android application as a client part and a Flask application as a server part.

Repository Structure

• mobile-app -- source code of an android mobile application
• neural-network -- source code for training and testing the neural network
• web-server -- source code of a flask web application
• papers -- pdf documents with project documentation
• app-preview.mov -- video-preview of a mobile application powered by android 8

Dependencies

$ pip install -r BeautyDeep/mobile-app/requirements.txt
$ pip install -r BeautyDeep/neural-network/requirements.txt
$ pip install -r BeautyDeep/web-server/requirements.txt

Usage

For local server

1. Install all dependencies for application and server
2. Connect your phone and computer devices to the same network
3. Open "BeautyDeep/web-server" and start the server.py
4. Wait until the neural network configurations are loaded
5. Install the mobile application or open "BeautyDeep/mobile-app" on your computer and run main.py
6. In the server settings of the application, specify the public ip address (e.g. http://192.168.0.102:5000) of your local machine

For remote server

1. Deploy the web application to the remote server or run setup.sh on your remote PowerShell
2. Install the mobile application or open "BeautyDeep/mobile-app" on your computer and run main.py
3. In the server settings of the application, specify the domain (e.g. http://domain.io) or public ip address (e.g. http://34.91.143.244:5000) of your remote server