Convolutional neural networks (CNNs) allow the computer to classify images. Apart from classifying objects they also are able to give us insights on what makes a picture. What is the essence of a picture? By visualizing the layers of CNN architectures we dive into the understanding of how machines process images. This gives provides also insights into how the human “sees” pictures.
The article shall on one side present what elements build a picture and also provide code for a Python implementation with Keras.
Table of Contents
- Disclaimer
- The base image
- Using a VGG network
- Convolutional Layer Feature Maps
- What can we observe?
- Visualize with code
- Features in grid
- Features with notebook display
- Bonus — a neural transfer approach
- Inspiration
- About
Disclaimer
I am not associated with any of the services I use in this article.
I do not consider myself an expert. If you have the feeling that I am missing important steps or neglected something, consider pointing it out in the comment section or get in touch with me.
I am always happy for constructive input and how to improve.
This was written on 25–10–2020. I cannot monitor all of my articles. There is a high probability that when you read this article the tips are outdated and the processes have changed.
If you need more information on certain parts, feel free to point it out in the comments.
The base image
Source https://unsplash.com/photos/hJKkyoG8_ng by Hadi Yazdi Aznaveh — Unsplash Award 2019 selected in “Current events”
I like this image because it has a story, which sparks emotion but also provides a rich structure on a photographic level.
#deep-learning #machine-learning #computer-vision #explainable-ai #convolution-neural-net
