Skip to content

Latest commit

 

History

History
28 lines (24 loc) · 951 Bytes

README.md

File metadata and controls

28 lines (24 loc) · 951 Bytes
Raw

Squad-CAM

A project based on further investigating the Grad-CAM explanation method.

Table of contents

General info

This project compares Squad-CAM with other explanation methods like Grad-CAM and guided backprop. The target is DCNN models like Resnet, VGG16 or AlexNet. This project uses VGG16 as a base for experiments. This requires all images to be of size 224x224. We opted not to resize images due to distortions.

Technologies

Project is created with:

  • Tensorflow version: 2.1.0
  • Keras version: 2.3.1

Setup

To run this project, set up a conda enviroment and run

$ pip install -r requirements.txt

Analyze a set of photos by defining the source and target folders

$ python Squad-CAM -source images/*.png -target output