The goal of plant breeders is to produce crops with beneficial phenotypes. Our lab is working to provide automated tools and algorithms to extract phenotypic traits from various imaging modalities including aerial drone imagery, satellite imagery, ground-based imagery and video, and neutron imaging tomography of plant root systems. This resarch is conducted within the Plant Phenotyping and and Imaging Research Centre (P2IRC) funded by a Canada First Research Excelence Fund from the Natural Sciences and Engineering Research Council of Canda.
Joint work with the Canadian Food Inspection Agency in which we seek to automatically identify and classify weed seed species. Applications include seed sample purity analysis, and phytosanitary certification. This research is funded in part by the Canadian Food Inspection Agency.
Working with difficult segmentation problems such as some of the problems above motivates the use of semiautomatic segmentation algorithms. We are particularly interested in algorithms where there is an interactive phase where an operator enters some high-level contextual information which is followed by a fully automatic phase. Most evaluations of such algorithms are performed by having a small number of human operators segment each image once, or just a few times. This limits the number of interactive inputs that are encountered by the algorithm, and therefore the confidence and accuracy of the evaluation. We are investigating methods of overcoming these difficulties. We are also interested in rigorously evaluating the relationships between segmentation accuracy, precision (reproducability), and operator time and the method by which contextual information is input (efficiency).
We were inspiried by histopathology wherein different kinds of cells are more easily distinguished by chemically staining thin sections of biopsy tissue mounted on microscopic slides. We have developed some methods to digitally "stain" image texture by modifying local texture as a function of itself. This may achieved without knowledge of where each kind of tissue appears in the image (ie. without prior segmentation). We anticipate applying our methods on medical images for clinical decision support and radiomics.
Much of my recent research has been centered around segmentation, detection, and classification applications for medical images. Recent and ongoing projects include:
We are a part of a consortium of research teams from the UK, Canada and the Netherlands investigating how we can use interactive systems design in conjunction with image processing and text mining techniques to help archaeologists find, organise and analyse the thousands of image and document resources available to them for answering archaeology research questions. Our team was funded the Social Sciences and Humanities Research Council of Canada via the Digging into Data Challenge program.
Our lab was pleased to partner with Seed Hawk Inc. to develop their SCT Savings Simulator™. The software is being used by Seed Hawk sales representatives to provide estimates to customers of potential savings from using Seed Hawk's Sectional Control Technology which are customized to the user's particular piece of land. The software analyzes a satallite photo of the customer's farmland, and computes a seeding strategy. Taking into account seed and fertilizer costs, it predicts an estimated cost savings that could be realized through the use of SCT.