ArtSEEker - using headless Drupal to power AI Art recognition
Time: Wednesday 20th March 10:15-10:45
Room: Main Ballroom
Track: Drupal Development
This talk will take the audience through a cutting-edge AI project, and how headless Drupal is a key component to delivering this app.
The Queensland Art Gallery | Gallery of Modern Art (QAGOMA) has embarked on an ambitious project: leveraging sophisticated AI mechanisms to make our digital content easily accessible on personal devices. Our product, entirely built on open-source tech and built in-house, employs computer vision to recognise artworks with high accuracy (>98%) and rapidity (averaging 200ms) across thousands of pieces. The Galley visitor can simply point their mobile phone at any work on display, and the AI will recognise the art object and return all relevant content, as well as inviting engagement and extra layers of meaning with the work. This negates the need for markers such as QR codes, label keys, or other intricate technological tools.
Central to our Web App's functionality is its computer vision recognition feature. By pointing a mobile device at an artwork—be it 2D or 3D—the artwork is identified within a few hundred of milliseconds to return related digital content, such as descriptions, colour and shape analyses, interactive components, and interactive questions.
In this presentation, I’ll guide attendees through our holistic journey, from the technological foundation to the pilot phase, taking insights from lessons learned and discussing our vision for a museum-wide rollout.
Importantly this talk will be anchored in what is providing the data – a Drupal 10 instance providing a custom data API, and many specific endpoints. During AI inference training this was delivering 100,000s of responses in a day and scaling appropriately.
The Queensland Art Gallery | Gallery of Modern Art (QAGOMA) has embarked on an ambitious project: leveraging sophisticated AI mechanisms to make our digital content easily accessible on personal devices. Our product, entirely built on open-source tech and built in-house, employs computer vision to recognise artworks with high accuracy (>98%) and rapidity (averaging 200ms) across thousands of pieces. The Galley visitor can simply point their mobile phone at any work on display, and the AI will recognise the art object and return all relevant content, as well as inviting engagement and extra layers of meaning with the work. This negates the need for markers such as QR codes, label keys, or other intricate technological tools.
Central to our Web App's functionality is its computer vision recognition feature. By pointing a mobile device at an artwork—be it 2D or 3D—the artwork is identified within a few hundred of milliseconds to return related digital content, such as descriptions, colour and shape analyses, interactive components, and interactive questions.
In this presentation, I’ll guide attendees through our holistic journey, from the technological foundation to the pilot phase, taking insights from lessons learned and discussing our vision for a museum-wide rollout.
Importantly this talk will be anchored in what is providing the data – a Drupal 10 instance providing a custom data API, and many specific endpoints. During AI inference training this was delivering 100,000s of responses in a day and scaling appropriately.
Speakers
Morgan Strong
Morgan is the digital transformation manager at the Queensland Art Gallery of Modern Art. Morgan has been working with technology in the GLAM sector for almost 15 years, collaborating with institutes across the country. Morgan's work focuses on leveraging technology to create meaningful, to create meaning from collection data and improve the way museums can embrace digital to improve the way they do their work.