The International Association of Geomorphologists (IAG) is celebrating #IntGeomorphWeek2021 from 1-7 March. On 2 March between 09:00 and 13:00 the IAG is celebrating Young Geomorhologists (or Emerging Career Researchers) and their work. Join Mr Barend van der Merwe, a lecturer in our department, on 2 March 11:00-11:20 local time (10:00-10:10 CET) for his talk on The Performance of Different CNN Architectures on Barchan Asymmetry. The talk will be given on Zoom and you can register at https://www.eventbrite.co.uk/o/international-association-of-geomorphologists-32264657625. Barend's abstract is given below.
Classification Convolutional Neural Networks (CNN's) promise to be a useful tool for automating the image interpretation workflow. As it result, it is possible to increase the amount of data derived from imagery for geomorphological research. Despite this, the application of CNNs to aeolian landforms have been minimal. A particular gap in the knowledge is the use of CNNs to classify barchans based on their morphometric characteristics (i.e. shape). We investigate the suitability of different CNN architectures and training parameters to binary asymmetry classification tasks that incrementally increase in difficulty. Preliminary results are presented.
If you would like to know more about the local chapter of Young Geomorphologists, please have a look at https://saageomorph.wixsite.com/saag/about-sayg. The current chair of the SAYG is Renee Grundling, a Masters student in our Department.