Ian Stavness from the University of Saskatchewan presented a webinar, "Low-cost field imaging
Using the ProTractor for inexpensive ground-based field imaging. (Photo credit: Cam Kenny, Agriculture and Agrifood Canada)
for plant phenotyping," on Wednesday, August 7, 2019 at 10:30 am CST (UTC -6).
The edited video is available through YouTube, viewing on our page here, and download as an .mp4 [195MB].
Talk details.
Title: Low-cost field imaging for plant phenotyping.
Abstract:Low-cost imaging sensors, such as consumer cameras, have the potential to democratize image-based field phenotyping by making image acquisition systems cheaper and easier to replicate. However, such cameras typically have poor fidelity geo-location information. While it may be possible to augment aerial and ground vehicles with accurate RTK-GPS add-on systems, this increases the cost and complexity of imaging setups. In this talk, I will present work on localizing plots and rows within field trials from aerial and ground images without geo-location
information. I will introduce our “ProTractor” platform, co-developed with Agriculture and Agri-Food Canada, which uses inexpensive action cameras on an existing tractor/sprayer vehicle, and the associated image processing steps that we use to localize and crop individual rows of plants from sequences of top-down images from the field. I will also discuss machine learning approaches we have been developing for estimating plant phenotypes from outdoor images.
UAV imaging of Dr. Curtis Pozniak’s wheat breeding trials at the University of Saskatchewan. (Photo credit: Dr. Steve Shirtliffe)
Supplementary Materials:
Ubbens, J. R., & Stavness, I. (2017). Deep plant phenomics: a deep learning platform for complex plant phenotyping tasks. Frontiers in plant science, 8, 1190. link.
Higgs, N., Leyeza, B., Ubbens, J., Kocur, J., van der Kamp, W., Cory, T., Eynck, C., Vail, S., Eramian, M., & Stavness, I. (2019). ProTractor: A Lightweight Ground Imaging and Analysis System for Early-Season Field Phenotyping. In Proceedings of the IEEE Conference on Computer Vision and Pattern
Recognition Workshops (pp. 0-10). link.
Biographical Information.
Ian Stavness is an Associate Professor of Computer Science
Ian Stavness, webinar speaker on August 7, 2019.
at the University of Saskatchewan. He leads the Deep Learning for Phenomics project at the Plant Phenotyping and Imaging Research Centre (p2irc.usask.ca) at the University of Saskatchewan. He also directs the Biological Imaging & Graphics (BIG) lab focused on 3D modeling, image analysis and deep learning for biological and biomedical applications. His research directions target the emerging field of computational agriculture. He completed a post-doc at Stanford University with the NIH Center for Biomedical Computation and his PhD at the University of British Columbia.