In modern Australia, there is a gap in understanding of Aboriginal history and culture. This gap in knowledge and understanding undermines the importance and meaning of Contemporary Aboriginal art in Australian society.
Professor Simoff and the project team took on the challenge to create a machine learning based, educational tool with web interface for use in places such as museums, secondary schools, universities and art galleries, in order to bridge the gap in knowledge, and educate people about Aboriginal art, while helping create a better understanding of Indigenous culture.
Solution Description
The solution is a responsive web-based application, available on all devices with a modern web browser. It is designed to give users the choice on how they learn about Contemporary Aboriginal Art.
At its core, the application allows users to upload an image of Aboriginal Art of their choice, and receive an interpretation about the region, style, colours and symbols. The application can currently detect two known Aboriginal symbols: “waterhole” and “person”.
Users can then provide feedback on their interpretations, rating each section and providing comments. This application has been created using supervised machine learning. This decision was made to eliminate cultural concerns raised during the discovery phase of this project. The application also features user registration & login, which allows the user to store their interpretations in the database for future reference. Users can export these interpretations as a PDF, in a catalog style view. Finally, the application requires the user to receive written permission from the Artist before uploading their image to the application. This consent file is stored in the Database and can be used as consent for any of the artist’s images.
Technologies
The solution consists of a three-tier architecture, with web, application and database servers deployed as separate Docker containers automatically using GitLab’s CI/CD pipelines. For the database and web application the solution uses the MEAN Stack (MongoDB, Express.js, Angular, and Node.js). The machine learning application is built using PyTorch in Python and served over an API using Flask. Currently the solution is hosted on a Virtual Machine, running Linux CentOS, on the Western Sydney University network.
Developed by PS2021
Aiden Harris
Baris Demirci
Phoebe Lilius