This video shows a convolutional neural network transferring seismic wave into P phase probability density function during the training process. The predict label looks like the original trace at first, drop down rapidly and slowly matches the training label.
SeisNN: Deep Learning Seismic Phase Auto-Picking Package
- The project goal is making a solid pipeline for seismic phase auto-picking deep learning research and production.
- With this package, researchers can focus on making their own models.
- Source code is available on Github.