Automatic 3D Salt Segmentation with Deep Learning

Research papers
Publication: Shi et al., 2019, SaltSeg: Automatic 3D salt segmentation using a deep convolutional neural network. 

The authors defined the problem in 3D as binary image segmentation and presented an efficient solution based on a convolutional neural network with an encoder-decoder architecture.
The network is trained by randomly extracting sub-volumes from an open 3D SEAM dataset. The salt prediction results are shown on both SEAM and F3-block seismic data.

The link for the LinkedIn post