Globus streamlines research process at Franklin and Marshall College

Franklin & Marshall College in Lancaster, Pennsylvania, is a small liberal arts college with a little over 1,800 students and just two full-time IT staff dedicated to helping research faculty. Looking for ways to assist research staff—as well as making them self-sufficient—was paramount given the limited staff.

Dan Ardia, Professor of Biology at Franklin & Marshall College, approached F&M’s Research Computing group with a problem. He and his students do field work where they set up trail cameras in multiple different locations across the county. Collectively, these cameras capture an extremely large number of images, amounting to over 40TB of data annually. These images need to be transferred to long-term storage for back-up and computational analysis. Doing that with tools like SCP just wasn’t feasible. While Dan had already installed Globus Connect Personal on his laptop, and was using Globus to transfer images, he found the process to be clunky and problematic. Tony Weaver, Research Computing Programmer, approached Dan about how to streamline his workflow. Tony had noted that the process was very time consuming, and required some knowledge of Linux; and due to a lack of undergraduate students’ familiarity with Linux, once the images were captured, Dan’s research students could not assist with the image analysis.

Dan and Tony explored alternative approaches to streamline and automate the data management tasks involved in his nature conservation research. After considering the multiple steps involved in uploading, archiving, and processing via machine learning field images, Tony created a Globus Flow for Dan that automates a two-step process. The first step is data transfer: Dan selects the source and destination for the data. The second step is computation for image analysis: Dan selects the location for the output of the analyses (a Globus endpoint and directory)—and he can also adjust a few other parameters for the image analysis software as needed. There are also some behind the scenes calculations within the flow to ensure the path information on the compute endpoint is correct based on the paths selected in the flow.

Using this simple yet powerful Globus flow, Dan has automated a big part of his research process, and is able to more efficiently process the images he and his students are capturing in the field. Now he can send his students into the field for data collection, and have them assist him with the image analysis.