Shape recognition of volcanic ash by simple convolutional neural network
This provides a method for volcanologists to automate shape analysis of tephra grains, potentially aiding eruption mechanism studies, though it is incremental as it adapts an existing CNN to a new domain.
The paper tackled the problem of volcanic ash shape recognition by applying a simple convolutional neural network (CNN) originally designed for handwritten digits to tephra images, achieving approximately 90% accuracy.
Shape analyses of tephra grains result in understanding eruption mechanism of volcanoes. However, we have to define and select parameter set such as convexity for the precise discrimination of tephra grains. Selection of the best parameter set for the recognition of tephra shapes is complicated. Actually, many shape parameters have been suggested. Recently, neural network has made a great success in the field of machine learning. Convolutional neural network can recognize the shape of images without human bias and shape parameters. We applied the simple convolutional neural network developed for the handwritten digits to the recognition of tephra shapes. The network was trained by Morphologi tephra images, and it can recognize the tephra shapes with approximately 90% of accuracy.