GitPlanet

Cheap and reliable Node.js hosting starts at $3/month, and $1/month static HTML hosting

Created with love in Canada, visit hostnodejs.com today

Feel like to post an Ad? Learn Details
All Projects → soskek → Dynamic_routing_between_capsules

soskek / Dynamic_routing_between_capsules

Licence: mit
Implementation of Dynamic Routing Between Capsules, Sara Sabour, Nicholas Frosst, Geoffrey E Hinton, NIPS 2017

Programming Languages

python
139335 projects - #7 most used programming language

Labels

cnnchainernips-2017capsnet

Projects that are alternatives of or similar to Dynamic routing between capsules

See
Code for the AAAI 2018 publication "SEE: Towards Semi-Supervised End-to-End Scene Text Recognition"
Stars: ✭ 545 (+169.8%)
Mutual labels:  cnn, chainer
Capsnet
CapsNet (Capsules Net) in Geoffrey E Hinton paper "Dynamic Routing Between Capsules" - State Of the Art
Stars: ✭ 423 (+109.41%)
Mutual labels:  chainer, capsnet
Lidc nodule detection
lidc nodule detection with CNN and LSTM network
Stars: ✭ 187 (-7.43%)
Mutual labels:  cnn
Sppnet Pytorch
A simple Spatial Pyramid Pooling layer which could be added in CNN
Stars: ✭ 198 (-1.98%)
Mutual labels:  cnn
Cnn Visualization
TensorFlow implementations of visualization of convolutional neural networks, such as Grad-Class Activation Mapping and guided back propagation
Stars: ✭ 194 (-3.96%)
Mutual labels:  cnn
Freihand
A dataset for estimation of hand pose and shape from single color images.
Stars: ✭ 187 (-7.43%)
Mutual labels:  cnn
Deep Learning With Python
Deep learning codes and projects using Python
Stars: ✭ 195 (-3.47%)
Mutual labels:  cnn
Tensorflow Mnist Cnn
MNIST classification using Convolutional NeuralNetwork. Various techniques such as data augmentation, dropout, batchnormalization, etc are implemented.
Stars: ✭ 182 (-9.9%)
Mutual labels:  cnn
Pratik Derin Ogrenme Uygulamalari
Çeşitli kütüphaneler kullanılarak Türkçe kod açıklamalarıyla TEMEL SEVİYEDE pratik derin öğrenme uygulamaları.
Stars: ✭ 200 (-0.99%)
Mutual labels:  cnn
Face Nn
游戏捏脸，基于神经风格迁移框架生成逼真人脸
Stars: ✭ 192 (-4.95%)
Mutual labels:  cnn
Deeptam
DeepTAM: Deep Tracking and Mapping https://lmb.informatik.uni-freiburg.de/people/zhouh/deeptam/
Stars: ✭ 198 (-1.98%)
Mutual labels:  cnn
Did Mdn
Density-aware Single Image De-raining using a Multi-stream Dense Network (CVPR 2018)
Stars: ✭ 192 (-4.95%)
Mutual labels:  cnn
Depression Detect
Predicting depression from acoustic features of speech using a Convolutional Neural Network.
Stars: ✭ 187 (-7.43%)
Mutual labels:  cnn
Lq Nets
LQ-Nets: Learned Quantization for Highly Accurate and Compact Deep Neural Networks
Stars: ✭ 195 (-3.47%)
Mutual labels:  cnn
Bilinear Cnn Tensorflow
This is an implementation of Bilinear CNN for fine grained visual recognition using TensorFlow.
Stars: ✭ 187 (-7.43%)
Mutual labels:  cnn
Cnn 3d Images Tensorflow
3D image classification using CNN (Convolutional Neural Network)
Stars: ✭ 199 (-1.49%)
Mutual labels:  cnn
Anime4k
A High-Quality Real Time Upscaler for Anime Video
Stars: ✭ 14,083 (+6871.78%)
Mutual labels:  cnn
Cnn Re Tf
Convolutional Neural Network for Multi-label Multi-instance Relation Extraction in Tensorflow
Stars: ✭ 190 (-5.94%)
Mutual labels:  cnn
Liteflownet2
A Lightweight Optical Flow CNN - Revisiting Data Fidelity and Regularization, TPAMI 2020
Stars: ✭ 195 (-3.47%)
Mutual labels:  cnn
Cyclegan Music Style Transfer
Symbolic Music Genre Transfer with CycleGAN
Stars: ✭ 201 (-0.5%)
Mutual labels:  cnn
View All Similar Projects ➔

Dynamic Routing Between Capsules

Chainer implementation of CapsNet for MNIST.

For the detail, see Dynamic Routing Between Capsules, Sara Sabour, Nicholas Frosst, Geoffrey E Hinton, NIPS 2017.

python -u train.py -g 0 --save saved_model --reconstruct

Test accuracy of a trained model (without reconstruction) reached 99.60%. The paper does not provide detailed information about initialization and optimization, so the performance might not reach that in the paper. For alleviating those issues, I replaced relu with leaky relu with a very small slope (0.05). The modified model achieved 99.66% (i.e. error rate is 0.34%), as the paper reported.

Visualization through Reconstruction

python visualize.py -g 0 --load saved_model

produces some images for analyzing digit capsules.

Different masks

vis_all.png

The top green images are real images which are given to the model. Blue images in i-th represents reconstructed ones of digit "i".

If an correct digit is selected as a target, the model reconstructs an image well (see the diagonal cells).

If an irrelevant target is selected, the reconstructed image gets spoiled (see "0" and the others in the column leftmost), maybe because of lack of information in its digit capsule. However, reconstruction toward a relevant target is not always spoiled, even if a target is not correct (see "8" and "9" the column rightmost).

Interpolation of values in digit capsules

Here, I show reconstructed images after linearly tweaking the value in a dimension in the capsule (as well as section 5.1 and figure 4 in the paper). Green images in the center are reconstructed images without perturbation. Note that a dimension has a different factor if the digit capsule differs, because each matrix for reconstructing each digit is unshared.

You can find and enjoy some factors of variation.

vis_tweaked0.png

vis_tweaked1.png

vis_tweaked2.png

vis_tweaked3.png

vis_tweaked4.png

vis_tweaked5.png

vis_tweaked6.png

vis_tweaked7.png

vis_tweaked8.png

vis_tweaked9.png

Note that the project description data, including the texts, logos, images, and/or trademarks, for each open source project belongs to its rightful owner. If you wish to add or remove any projects, please contact us at [email protected].