Project files and code used for research in fitting dispersion curve images, then using the fit images to compute seismic velocity in the area as well as building codes. The first neural network generates dispersion curve images using a GAN, since we currently do not have the requires amount of data to train an image segmentation neural network. The second neural network is the architecture of the image segmentation convolutional neural network used to classify the different portions of the dispersion images and separate out the different frequency vs velocity amplitude bands. Once the amplitude bands are separated and the underlying data read, we can then cluster the data then fit the data before inversion.
avand56 / dispersion-curve-gan-and-image-segmentation-cnn Goto Github PK
View Code? Open in Web Editor NEWProject files and code used for research in generating dispersion curve data using a GAN, then using the generated GAN data to train an image segmentation CNN