3DGradCAM

This is a repository for 3D Gradient weighted Class Activation Maps for Manufacturability Analysis. For more details, please see the paper: Learning Localized Geometric Features Using 3D-CNN: An Application to Manufacturability Analysis of Drilled Holes
Aditya Balu, Sambit Ghadai, Kin Gwn Lore, Gavin Young, Adarsh Krishnamurthy, Soumik Sarkar

Packages needed

Keras 2.0.5, Tensorflow 1.2

You can either import the GradCAM function:

from GradCAM import prepareGradCAM, GradCAM
activationFunction=prepareGradCAM(cnnModel, layerIdx, nbClasses)

activationFunction is a Keras Function for computing the class activation map

BSD

The following codes shall be made available soon:

Data Generation using ACIS solid modeling kernel
GPU Accelerated Voxelization: Available now @ https://github.com/idealab-isu/GPView
Training
GPU accelerated Volume Rendering of GradCAM output