Huadong Liao

I didn't encounter such problem. Could you tell me these: 1. except for the data reading pipeline, did you modify the code and which part; 2. Is the number of...

The code looks fine. I'm not sure what the problem is. What I can guess is that capsnet might have bias problem. But before making this conclusion, I suggest: 1....

You do not understand squashing correctly. Squashing ensures the length (euclidean norm) of its output vector is in [0, 1], not about the element in the vector. Have fun with...

`tf.summary.image` can deal with this problem and display the images correctly. It's OK since we just want to display it on `tensorboard`.

The length of vector in the CapsNet paper represents the [Euclidean norm](https://en.wikipedia.org/wiki/Norm_(mathematics)), which a.k.a [Euclidean distance](https://en.wikipedia.org/wiki/Euclidean_distance) of vector, in TensorFlow, you can simply call [`tf.norm`](https://www.tensorflow.org/api_docs/python/tf/norm) to calculate the length of...

Yes, I created a new project to meet your situation. Please see my [CapsLayer](https://github.com/naturomics/CapsLayer) repo. You can `import capslayer` and then call it in your code: `capslayer.layers.conv2d(...)` or `capslayer.layers.fully_connected(...)`

python can not find your dataloader, you should import your module to this [__init__.py](https://github.com/naturomics/CapsLayer/blob/master/capslayer/data/datasets/__init__.py) file

I tested it on MNIST and CIFAR10, but not smallNORB since I found that matrix capsnet is worse than dynamic routing capsnet: 1. matrix capsnet is easy to overfit. training...

I created a tool to visualize the predicted labels. you can use it directly. The idea of that tool is get all the predicted labels(10 labels) for each sample, then...

Please read the README carefully! >If you need to apply CapsNet model to your own datasets or build up a new model with the basic block of CapsNet, please follow...