Monday, April 8, 2019 (Introduction)¶

Attendants: Ulf, Axel, Richard, Malin, Luana, Maren, Josefine, Katharina, Sophia, Subir, Thomas

short introduction session, Sophia introduced the project, showing some pictures of what the data would look like
collected some questions regarding the data (see document: questions.md)
Sophia will visit her family on Wednesday. Malin, Richard and Thomas might go as well to have a look at the machine and get answers to questions
Harvest season lasts until June 24th, we want to gather as much data as possible, collecting it on a harddrive (possibly rotating multiple harddrives) to later store it on the university’s server -> Ulf
we are expecting 500 GB of image data throughout the entire season (guesstimated from 10 tons of asparagus at 100 grams each and 5 MB for each picture)
we will write an overview over the different classes, with a description for each class -> Sophia
participants of this study project will probably have to attend another seminar, presenting progress on the project every now and then
one final report will be written in the end, roughly 10 pages per participant with equal workload, ideally working code that interacts with the machine

Current Plan¶

first, we’ll acquire as much labelled data as possible, we will see how detailed the labels can be
set up preprocessing pipeline (cutting images into three parts)
train a classifier on images, two approaches:
- try to extract features like width and color
- try to train CNN or similar architecture on images themselves
interfacing with the machine itself is a second step