Dairy animal udder shape modeling - BC-389

Preferred Disciplines: Computer Engineering/Science, Medical Imaging (Masters, PhD, Post-Doc)
Project length: 8-12 months (2 units)
Approx. start date: As soon as possible
Location: Vancouver, BC
No. of Positions: 1
Preferences: UBC / SFU
Company: EIO Diagnostics

About Company:

EIO Diagnostics combines multispectral imaging with machine learning to the problem of early detection of dairy animal udder infections (mastitis). 

Summary of Project:

EIO Diagnostics combines multispectral imaging with machine learning to the problem of early detection of dairy animal udder infections (mastitis). Dairy animals such as cows and goats have variable udder shape created by a variety of factors such as stage of lactation, genetics, age, and past infections.

In a dairy, the udders can be imaged from the side and posterior of the animal with various advantages and disadvantages for the detection of infection which is further compounded by the udder shape. The illumination of an udder in a milking parlour tends to be poor as the udder is under the animal and often further shadowed by milking equipment.

This project aims to “normalize” the shape of an animal’s udder across milking and between animals by geometrically modeling the udder shape from images (visible and infrared). This will allow the warmer areas created by denser connective tissues and contact between the legs and the udder to be distinguished from infected tissues. It is also hoped that by comparing the shape between milking that edema caused by fluids collecting in the udder unrelated to lactation can be detected.

Additionally, the geometric model will allow the image pixel intensities and areas to be corrected for the

  • Radiation effects caused by the angle between the surface normal and the camera view axis,
  • Camera perspective distortion 

Background and required skills

Research Objectives/Sub-Objectives:

  • Literature search to establish methods and materials
  • Implementation of modeling method
  • Comparison of rendered geometric models to alternative images

Methodology:

    • TBD

    Expertise and Skills Needed:

    • Computer graphics / computer vision experience with geometric 3D models and the fitting of the models to images
    • Software development in C, C++, Matlab or other agreed upon programming language

    For more info or to apply to this applied research position, please

    1. Check your eligibility and find more information about open projects.
    2. Complete this webform. You will be asked to upload your CV. Remember to indicate the title of the project(s) you are interested in and obtain your professor’s approval to proceed!
    Program: