| Abstract Detail
Recent Topics Posters Weaver, William [1], Ng, Julienne [2], Laport, Robert [3]. Rise of the machines: does plant blindness affect machine learning applications, too? Optimizing biological data generation and analyses is a perennial concern among biologists working with large datasets, limited time, and budgetary constraints. Automating repetitive processes could greatly enhance science output, enriching our understanding of the living world, but is complicated by the inherent variance and complexity of biological systems. Recentadvances in the fields of machine learning and computer vision promise an array of tools capable of coping with biological complexity and producing exceptional datasets. For example, applications that automaterelevant data procurement from millions of digitized herbarium specimens could open the door to ground-breaking ecological, genetic, evolutionary, and climate change research. We have developed the software package, LeafMachine, to automate the process of measuring leaf traits on digitized herbarium specimens. Yet, many challenges remain for data procurement automation. Chief among these are evaluating precision and accuracy of automated measurements, designing for generalizability, and maintaining an appreciation of the idiosyncrasies of biological systems. Here we present and discuss challenges encountered and lessons learned from our application of computer vision and machine learning to automate phenotypic trait data extraction from digitized herbarium specimens, with an eye toward driving a dialogue on establishing best practices for the future proliferation of such approaches in the plant sciences. Log in to add this item to your schedule
1 - University Of Colorado Boulder, Department Of Ecology & Evolutionary Biology, Campus Box 334, Boulder, CO, 80309, USA 2 - University Of Colorado Boulder, Department Of Ecology & Evolutionary Biology, Campus Box 334, Boulder, CO, 80309, United States 3 - Rhodes College, Department Of Biology, 2000 North Parkway, Memphis, TN, 38112, United States
Keywords: machine learning convolutional neural networks support vector machines herbarium specimen images automation morphology.
Presentation Type: Session: P, Recent Topics Posters Location: Arizona Ballroom/Starr Pass Date: Monday, July 29th, 2019 Time: 5:30 PM Number: PRT011 Abstract ID:1378 Candidate for Awards:None |