Friday, May 10, 2019 (put topic here)

Attendants: Ulf, Axel, Malin, Luana, Michael, Josefine, Sophia, Thomas, Maren

Summary about visit to Rheine

(Sophia and Thomas went)

  • Collected labeled data

  • measured distance camera to band (~60 cm), and asparagus width

  • service to move images to harddrive works

Problems

  • Service to move the images leads to an error that stops the machine

  • to upload pictures directly to the university server is limited by speed of GPU, Internet, or external harddrive

  • compressing the images to jpag rather than zip could be faster

To our work

  • we should start compressing the pictures

  • preprocessed images are sufficently smaller

The Hermelerguy

  • he is interested in a cooperation with us (could provide a newer machine and camera) - “feels like a profit making person”

  • How could we benefit from the Hermelerguy?

    • probably he has labeled data, that he is not willing to give without cooperation

    • a lot of knowledge how the machine works

    • get some alternative data, that could improve our results (2 datasets) or get some new data, to make some variance (could maybe then be used for other machines?)

    • get the API (holy grail) of the machine

    • we want to improve the old machine, rather than a new one

Organisatorial

  • internal work party at the asparagusfarm at the 18.05.19, 7 pm (BBQ and drinks are provided)

  • We can use the room, where we met the first week (50/228 ?), we are going to apply for a transponder but for now we can just ask Ulf/Axel to open

  • We will (try) start using Asana from now on

Next steps

  • Prepare a meeting with the Hermelerguy:

    • possible invitation 24.05.19

    • presentation: who are we, what are we doing, what did we already do, what do we want to archive (slides: https://docs.google.com/presentation/d/1EO0M33C2WHCRcTHvgkIY2ezVKcoGBmP9NbNafQ2zRKk/edit?usp=sharing)

    • collect questions

    • collect topics, that we avoid!!!1elf!!

    • contact the software company, to get connected

  • start compressing the pictures

For next meeting (17.05.19)

  • Everybody should present one classification architecture in the Deep learning class (sign up in Asana, presentation: https://docs.google.com/presentation/d/1fsgr7jhZ5YtJWwOv8TSN4qxvsZwf2KRwoQQdKJv-gnM/edit?usp=sharing)

  • how they are related to our task (qualification/disqualification)

  • plan another trip to Rheine to collect labeled data