My Bachelor's Thesis. Capsule Network Knowledge Extractor is a software to easily produce network visualizations of a Capsule Network created with Keras under TensorFlow 2.0.
The project requires Python 3 installed on your system. Install the dependencies:
pip3 install -r requirements.txt
Run a local server:
python3 run.py
Finally, navigate to 127.0.0.1:5000 to use the software.
Additionally, you can edit the Capsule Network model or re-train the current one. For more informations about that, please check ./capsnet_trainer/.
Convolutional layer's feature maps.
Primary Capsules' entities localization.
Improved Routing Path Visualization Technique.
Dense Capsules' magnitude and dimensions manipulated and visualized through the Decoder.
Area of interests produced by Grad-CAM++.
You can download or view the .pdf file of the thesis here. Please note that this file will never be updated, nor the latex files will be shared.
This project is licensed under the MIT License - see the LICENSE file for details.