Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Phenotyping of apples with a multi camera system (Master thesis or student project)
Apple breeding is intensively done in Switzerland, however phenotyping of the fruits is often done visually.
Therefore a Vision method which extracts objective values for shape, size and color should be established in this project.
Keywords: camera calibration, shape and size estimation of small objects in a fixed camera system, color analysis, fruit phenotyping
**Background**
With sales of more than 200
million CHF per year apple is the most important fruit in Switzerland. Apple breeding is an important
part and intensively done. However, phenotyping of the fruits, is surprisingly still often done manually,
which makes results and scorings of different persons hard to compare.
Vision methods are well suited to get objective values for shape, size and color.
**You will** optimize the current setup of 5 cameras, integrate a fluent acquisition process which
immediately uses a multi camera approach to extract shape, size and color of a single apple and saves the
results. A pre setup of the system already exists which will enable you to start your work rapidly. The
aim is to build an effective tool for visual apple phenotyping in an interdisciplinary work.
- Optimize and calibrate the multi camera setup
- Extract shape and size of the fruit
- Create a surface color map of fruit skin (Gall–Peters projection)
- Color analysis of fruit skin
**Background**
With sales of more than 200 million CHF per year apple is the most important fruit in Switzerland. Apple breeding is an important part and intensively done. However, phenotyping of the fruits, is surprisingly still often done manually, which makes results and scorings of different persons hard to compare. Vision methods are well suited to get objective values for shape, size and color.
**You will** optimize the current setup of 5 cameras, integrate a fluent acquisition process which immediately uses a multi camera approach to extract shape, size and color of a single apple and saves the results. A pre setup of the system already exists which will enable you to start your work rapidly. The aim is to build an effective tool for visual apple phenotyping in an interdisciplinary work.
- Optimize and calibrate the multi camera setup - Extract shape and size of the fruit - Create a surface color map of fruit skin (Gall–Peters projection) - Color analysis of fruit skin
- Clean and commented Matlab code for data acquisition, robust shape and size extraction of apples, creation of a color map of the apple surface
- Validation of the developed method with a set of apples including an analysis of the accuracy
- of the method. Apples will be provided for this task
- A basic description of the workflow should be provided
**The ideal candidate** is highly motivated and would have:
- a background in computer science, computer vision or similar (and knows how to program)
- an interest in working with applied topics in plant context
- some insight into stereo or multi camera systems would be an advantage
The master thesis will be carried out in a collaboration between Agroscope and the Crop Science Group
of ETH Zurich.
- Clean and commented Matlab code for data acquisition, robust shape and size extraction of apples, creation of a color map of the apple surface - Validation of the developed method with a set of apples including an analysis of the accuracy - of the method. Apples will be provided for this task - A basic description of the workflow should be provided
**The ideal candidate** is highly motivated and would have:
- a background in computer science, computer vision or similar (and knows how to program) - an interest in working with applied topics in plant context - some insight into stereo or multi camera systems would be an advantage The master thesis will be carried out in a collaboration between Agroscope and the Crop Science Group of ETH Zurich.
For any questions or details, please contact Dr. Norbert Kirchgessner (main supervisor of the
thesis) norbert.kirchgessner@usys.ethz.ch
For any questions or details, please contact Dr. Norbert Kirchgessner (main supervisor of the thesis) norbert.kirchgessner@usys.ethz.ch