Researchers from Cardiff University are using artificial intelligence (AI) to automatically segment and capture information from museum specimens and perform data quality improvement without human input.
The university has been working with museums from across Europe including the Natural History Museum, London. The AI is being used to refine and validate new methods and contribute to the mammoth task of digitizing hundreds of millions of specimens.
There are more than 3 billion biological and geological specimens in natural history museums globally. Digitizing these specimens — where the physical information is transformed into a digital format — has become a new task for museums as the digital world become ubiquitous.
The digitalization helps reduce the amount of manual handling of specimens, which are delicate and prone to damage. Having the data and images in a digital format can reduce the risk to the physical collection and protect specimens for the next few decades.
"This new approach could transform our digitization workflows,” said Laurence Livermore, deputy digital program manager at the Natural History Museum, London.
Image segmentation
Researchers tested a new method called image segmentation, where AI easily and automatically locates and bounds different visual regions on images as diverse as microscope slides or herbarium sheets with high accuracy.
This automatic segmentation can be used to focus on capturing specific regions of a slide or sheet. It can also help to perform quality control on the images to ensure that digital copies of specimens are accurate.
“In the past, our digitization has been limited by the rate at which we can manually check, extract, and interpret data from our images,” Livermore said. “This new approach would allow us to scale up some of the slowest parts of our digitization workflows and make crucial data more readily available to climate change and biodiversity researchers.”
The AI method was then trained and tested on thousands of images of microscope slides and herbarium sheets from different history collections including the Natural History Museum, Royal Botanic Gardens, Kew and Naturalis Biodiversity Center, Muséum National d'Histoire Naturelle, Museum für Naturkunde, Finnish Museum of Natural History, Meise Botanic Garden and Naturalis Biodiversity Center.
The full research can be found in the journal Machine Vision and Applications.
