Key points are not available for this paper at this time.
In response to the COVID-19 pandemic, public spaces such as museums and art galleries are experiencing increased demands to offer virtual online access. While current solutions seek to replace or augment a real visit, online tours often suffer from being too passive and lack in-depth interactivity to keep virtual visitors meaningfully engaged with an exhibition. Museums and art galleries seeking to broaden and engage their audience more deeply should offer intriguing experiences that invite the visitor to explore, to be entertained, and to learn by interacting with the content. We propose a novel virtual museum experience that utilizes multiple visualizations to contextualize a gallery’s digitized artworks with related artworks from large image archives. We make use of the WikiArt dataset that includes more than 200,000 images and offers diverse metadata used for comparative visual exploration. In addition, we apply machine learning methods to extract multifaceted information about the objects detected in the images and to compute similarities across them. Visitors of our virtual museum can interactively explore the artworks using different search filters such as artist, style, or object classes detected within an image. The results are displayed through interactive visualizations offering different perspectives on artwork collections, leading to serendipitous discoveries and stimulating new insights. The utility of our concept was confirmed by an informal evaluation with virtual museum visitors.
Meinecke et al. (Mon,) studied this question.