The Art Museum in Vejle Municipality has implemented a smart city-solution, with the overall purpose of obtaining knowledge about the museum visitors and the way they use the museum. Such knowledge will help the museum in analyzing its audience and optimize the displays and exhibitions.
Previously, the Art Museum in Vejle has used common “visitor counters” as well as manual counts to gain insight into the use and users of the museum. With the implementation of new innovative AI technology, much more extensive data about the users (e.g. age, gender, and mood) and use (e.g. heatmaps and flow visualizations) of the museum can be collected and utilized, laying the foundation for better management information and a more data-driven decision-making process. The uses are many, but some issues that the art museum is looking to gain more insight into are:
-
Target groups: Are the exhibitions reaching the expected target groups, or should some other efforts be done in order to actively draw in specific segments of the population?
-
Flow: How is the flow of people in the museum? Are all corners of the museum used, or do the visitors move around in certain (less than ideal) patterns?
-
Popularity of certain spaces and works of art: Are some rooms more “popular” than others, and how is the visibility of the artworks in e.g. hallways increased? Are some works of art too “popular” for a certain space (should a very popular work of art be moved to a bigger space)?
-
Mood: What impact does a given work of art have on the mood of the visitor? Are the artworks perceived as interesting, beautiful or boring?
Applying multiple technologies in one solution, and building on standard algorithms
Besides the knowledge and insights that the gathered data can help create, another purpose of the project has also been to experiment with integrating different technologies into one solution. Thus, the project showcases how it is possible to experiment with the interplay between Internet-of-Things (IoT), Machine Learning (ML) and Artificial Intelligence (AI). The solution makes use of cameras as IoT-devices, ML to detect whether a person is in the image-feed and AI to detect a person’s characteristics.
Based on the non-functional requirements, that was made before the implementation, the solution is using Microsoft Azure’s Cognitive Services algorithms and Face API to detect the physical characteristics of a person. Using standard algorithms instead of individually developed ones has been a priority from the beginning of the project, since it makes it easier to ensure the further development of the algorithms and makes it possible to shift vendor to another Microsoft partner – if needed sometime in the future - without interfering with the algorithms.
Anonymization – protecting the identity of a person
The project takes several precautions to protect the identity of citizens who visits the art museum. Firstly, no biometric data or biometric templates are used to uniquely identify a person. Instead, the cameras are used to detect the physical characteristics of a person – only to classify some fundamental data about the person such as age, gender and perceived mood. Secondly, it must be highlighted that historical images are not saved from the cameras. Once the data from the camera has been processed – which takes a couple of seconds – the images are automatically deleted.
When it comes to visualization of the gathered data, this is all done by internal BI-consultants in the Culture-administration of the municipality using PowerBI. There is also a web application that employees at the art museum can get access to see some of the data.
For more information about the project, please contact Nicklas Fledelius at NIKFL@vejle.dk