Guest Blog by Phaedra Kokkini
If you visit our Digitisation Team, you might be drawn to one of our more curious imaging setups, the ‘Angled Label Image Capture and Extraction’ or to close friends: ALICE.
ALICE came about through the collaboration of Small Orders Curator Dr Ben Price and Digital Collections Project Manager Dr Steen Dupont in order to automate some of the processes to speed up our pinned insect digitisation.
A peculiar setup that used to feature a bucket and six cameras is now a hexagonal light box with cameras that can image pinned insect specimens at multiple angles to digitally extract the attached labels and could provide a breakthrough to an even faster specimen digitisation.
Dr Steen Dupont describes ALICE
I was very excited to start this project as the first test of ALICE’s true potential. What I didn’t expect was that the collection chosen to test ALICE would reveal some stunning wasps and an intriguing untold story of its previous owner.
When digitising pinned insect specimens, a digitiser usually has to take the specimen from the drawer, carefully remove its labels from the pin, place the specimen and the labels on a stage and after taking the photo repeat this process backwards with the same care. But what if we didn’t have to remove the labels from the pin of the specimen? This would speed up the imaging process, but would require a new way of extracting the data from the labels.
When using ALICE the only thing a digitiser needs to do is to place the specimen in the middle of the lightbox and hit ‘Capture’. ALICE takes six images simultaneously: a dorsal (top), a lateral (side) and four peripheral ones. The dorsal and the lateral images are useful views of the specimen. The peripheral views are there to capture a view of the nearest visible side of the upper label, which is usually the one that contains the most important metadata, such as the location, the collection date and the name of the collector.
Our Informatics Team and partners have developed software, that manipulates the label views of these images and combines them to digitally remove the pin. The label images can be used to transcribe the data of the labels as usual, to complete the digitisation process and can potentially also be used for automated data extraction such as Optical Character Recognition (OCR).
This is bypassing the step of removing the labels of a specimen for imaging, which is the part that takes the most time. In other pinned insect projects we have been able to image a maximum of 200-300 per person per day, but with ALICE we can get through 800 in less than a day, with a highest record of 1120 in a day! Moreover, the fact that we don’t need to manipulate the specimens to remove their labels means that this workflow minimises the possibility of causing damage to the collection.
The opportunity to test our new workflow came when the Museum acquired the insect collection of the late Martin Cooper containing an estimated 25,000 specimens. A large majority are Hymenoptera (bees, wasps and ants) along with some other insect orders. The specimens were to be rehoused, databased and incorporated in the Museum’s main collection. Cooper’s specimens were perfect for our pilot study because they were in very good condition. The data accompanying them was also good with most determined to species or genus level and with all data printed or clearly written on one label.
Before imaging and rehousing this collection, it was best practice to image the contents of every box as they came to the Museum for future reference. This gave an overview of the size and scale of the collection and allowed more accurate planning . For this task I used our SatScan machine to scan the 74 acquired insect collection boxes.
This was the start of what I can only describe as a detective story uncovering unexpected gains in digitisation and entomology. Stay tuned for the next part which will explore more about Martin Cooper and his collection.
To be continued…