Mapping Flowers to Save the Bees

Honey bees and native bees such as bumble bees are important pollinators in both natural and agricultural ecosystems. However, bees of both kinds are disappearing, and one of the key factors implicated in their decline is changes in land use which limit both the quantity and diversity of floral resources available to them. Traditional methods for assessing where flowers are tend to be either too labor-intensive to carry out on a large scale (ground-based surveys) or lack sufficient detail to discern individual flowering plants (satellite imaging). Instead, we are using high-resolution drone-based imaging methods, which provide greater detail than satellite images but scale up more easily than ground-based surveys. In this project, the student will continue developing and testing an existing method using computer vision and machine learning to create high-resolution maps of patches of flowers at a scale relevant to foraging bees.

In previous years, we have used a drone to collect detailed overhead photos of flowering plants within the Bernard Field Station, and recorded bee visitation at some of these plants. We have developed a software pipeline which stitches together those photos into an orthomosaic, then efficiently and accurately segments the images into plants vs. background, and classifies the plants as flowering buckwheat plants or something else. We would now like to quantify the floral resources available in a specific area, by (1) estimating the density and/or the flowering status of flowers on a buckwheat plant, and (2) mapping those density estimates back onto the orthomosaic. This will allow us to estimate the relative value and attractiveness of different areas to foraging honey bees.

Name of research group, project, or lab
The HMC Bee Lab
Why join this research group or lab?

You will be part of a team of students working on a set of related interdisciplinary projects, using mathematics, computation and engineering to solve problems of biological interest. The variety of techniques and approaches will give you an opportunity to explore your interests and develop new skills. This project, when completed, could contribute to pollinator health by helping researchers assess the quality of a habitat for bees. There may be opportunities to continue the work in a senior thesis, present at a regional or national conference, and/or co-author future publications.

Logistics Information:
Project categories
Biology
Computer Science
Computer Vision
Data Science
Natural Resources and Conservation
Student ranks applicable
Sophomore
Junior
Senior
Student qualifications

Strong programming experience in Python is required.

Here is a list of other skills/interests that could be useful, or which you might develop along the way. They're not requirements, but if you already have experience or interest in any of them, be sure to mention it in your essay.

  • Linux/UNIX command line and shell scripting, high performance computing
  • Computer vision and image processing, e.g. using OpenCV
  • Machine learning and classification
  • Data science and visualization, particularly with R
  • Environmental sustainability and ecology
  • Drone piloting
Time commitment
Fall - Part Time
Compensation
Academic Credit
Number of openings
2
Techniques learned

In this project, you will

  • contribute to a software pipeline combining Python and R
  • train, validate and tune a machine learning model
  • extract, analyze and visualize data resulting from the pipeline

You will also learn to read and discuss scientific literature, and to communicate across disciplinary boundaries and with the public about your work.

Contact Information:
Mentor
Matina Donaldson-Matasci
mdonaldsonmatasci@hmc.edu
Associate Professor of Biology
Name of project director or principal investigator
Matina Donaldson-Matasci
Email address of project director or principal investigator
mdonaldsonmatasci@g.hmc.edu
2 sp. | 5 appl.
Hours per week
Fall - Part Time
Project categories
Computer Vision (+4)
BiologyComputer ScienceComputer VisionData ScienceNatural Resources and Conservation