LightningBug: Mass Digitization of Pinned Insect Specimens
What: Overview of LightningBug project followed by open discussion
When: Thursday, August 15, 2019 - 1:00pm to 2:00pm EDT
Where: http://idigbio.adobeconnect.com/room/ select “Enter as guest,” enter your full name, and click Enter room.
LightningBug has the capacity to increase label transcription from the current rate of 2 million per year to tens of millions of transcribed labels produced per year. It will exponentially increase the number of specimen images produced per year and greatly increase research potential through 3-D rendering. LightningBug integrates two ongoing efforts, a label imaging processing (Argonne National Laboratory) and 3D specimen imaging system (COPIS –Yale). A synopsis of these efforts will be provided by Mark Hereld and Nicola Ferrier (Argonne), and Nelson Rios and Larry Gall (Yale). We welcome feedback and collaboration from the community as part of a process to finalize the proposal.
Multi-view imaging is the process of capturing series of images from varying viewpoints around an object. The resulting image sets can be used to deduce the three dimensional structure of photographed objects with high color fidelity. When applied to pinned insects, multi-view imaging not only permits the 3D reconstruction of specimens, but also the attached labels. We will discuss a proposed infrastructure for robotically-assisted multi-view imaging of pinned insects.
Label reconstruction proceeds by identifying fragmentary label views found in each image, triangulating three-dimensional locations of features on each label, and projecting these image fragments into a synthesized two-dimensional face-on view of the written content. Overcoming the occlusions from the specimen, pin, and other labels found in any single view, the virtual labels to be generated from these multi-image snapshots are then ready for transcription. Rapid progress in automated offline transcription of typewritten, hand-written, and mixed fields of text will enable inexpensive conversion of these virtual labels into digitally available data.
Need for LightningBug
- There are 300 million arthropod specimens in North American collections, only 5% have had labels transcribed and less than 1% have been imaged.
- These data are valuable and unique, for example 80 million “historic” specimens can serve as reference points for global change impacts but only two million have been digitized.
- Only technological breakthroughs will allow us to digitize these specimens in less than a lifetime. An NSF commitment to promote technologies that solve real-world problems and emerging technological advances make it possible to provide solutions that will exponentially increase our ability to transcribe labels and image specimens.
- The NSF-ADBC program has created a “digitization culture” among arthropod collections and a growing network of participating arthropod collections (over 100) can help guide developments and ensure that technologies will be implemented effectively across a wide spectrum of collections.
- We need to integrate developing technological advances to further engage the digitization community. This is especially true for the largest collections that curate most of the accumulated holdings.
This meeting will be recorded and posted on https://www.idigbio.org/wiki/index.php/SWG_Webinar_Series
Before the meeting please test your connection: https://admin.acrobat.com/common/help/en/support/meeting_test.htm
Get a quick Adobe Connect overview: http://www.adobe.com/products/adobeconnect.html
Please enter the Adobe Connect room 15 minutes before the meeting to test your mic and cam, then you can come back at the start of the meeting.
