GitHub - JeanElsner/capsules-neurorobotics: Evaluates the robustness of capsule networks in regard to novel viewpoints, occlusion and more

This repository includes several neural network models, that are evaluated based on the Visual Pattern Recognition (VPR) dataset. The VPR dataset is derived from a Neurorobotics Platform experiment, that creates sensor data for object recognition. This includes data augmentation scripts to test robustness to occlusion and novel viewpoints as well as a part-whole consistency.

Name		Name	Last commit message	Last commit date
Latest commit History 60 Commits
datasets		datasets
loss		loss
model		model
training		training
utils		utils
.gitignore		.gitignore
.gitmodules		.gitmodules
LICENSE		LICENSE
README.md		README.md
backprop_test.py		backprop_test.py
diehl_and_cook.py		diehl_and_cook.py
grid_seach.sh		grid_seach.sh
pytorch_grid_search.sh		pytorch_grid_search.sh
snn_backprop.py		snn_backprop.py
snn_reservoir.py		snn_reservoir.py
snn_supervised.py		snn_supervised.py
test.py		test.py
train.py		train.py
visualization.ipynb		visualization.ipynb

License

JeanElsner/capsules-neurorobotics

Folders and files

Latest commit

History

Repository files navigation

About

Topics

Resources

License

Stars

Watchers

Forks

Languages