# convnet-interpretability-keras

**Repository Path**: martbox/convnet-interpretability-keras

## Basic Information

- **Project Name**: convnet-interpretability-keras
- **Description**: Visualizing VGG16 Convolutional Neural Network using Keras
- **Primary Language**: Unknown
- **License**: MIT
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2020-01-28
- **Last Updated**: 2020-12-19

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Visual intrepretability for Convolutional Neural Network in Keras
Following visualization techniques are used:
* Visualizing intermediate activations (convolution outputs)
* Visualizing convolutional filter/kernels
* Visualizing input pixel space from intermediate activation using deconvnets
* Visualizing areas of interest in input image for each layer using GradCAM technique

Read more about this project in my blog [here](https://towardsdatascience.com/visual-interpretability-for-convolutional-neural-networks-2453856210ce)