Unsupervised classification of snowflake images using a generative adversarial network and <i>K</i>-medoids classification

Leinonen, Jussi; Berne, Alexis

doi:https://doi.org/10.5194/amt-13-2949-2020

Articles | Volume 13, issue 6

https://doi.org/10.5194/amt-13-2949-2020

© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/amt-13-2949-2020

© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 13, issue 6

Research article

|

05 Jun 2020

Research article |

| 05 Jun 2020

Unsupervised classification of snowflake images using a generative adversarial network and K-medoids classification

Jussi Leinonen and Alexis Berne

Data sets

Snow-gan-classification J. Leinonen https://github.com/jleinonen/snow-gan-classification

Model code and software

Snow-gan-classification J. Leinonen https://github.com/jleinonen/snow-gan-classification

Short summary

The appearance of snowflakes provides a signature of the atmospheric processes that created them. To get this information from large numbers of snowflake images, automated analysis using computer image recognition is needed. In this work, we use a neural network that learns the structure of the snowflake images to divide a snowflake dataset into classes corresponding to different sizes and structures. Unlike with most comparable methods, only minimal input from a human expert is needed.