From: A state-of-the-art technique to perform cloud-based semantic segmentation using deep learning 3D U-Net architecture
Patch size
64 × 64 × 64 voxels
Patch per image
16
Mini-batch size
8
