uark-aicv / 3dconvcaps Goto Github PK
View Code? Open in Web Editor NEW[ICPR 2022] 3DConvCaps: 3DUnet with Convolutional Capsule Encoder for Medical Image Segmentation
Home Page: https://arxiv.org/abs/2205.09299
License: Apache License 2.0
[ICPR 2022] 3DConvCaps: 3DUnet with Convolutional Capsule Encoder for Medical Image Segmentation
Home Page: https://arxiv.org/abs/2205.09299
License: Apache License 2.0
Hi!
According to the paper, patch_size is 323232, but bash file shows 646464, is the parameter in bash file the optimal value?
Hi!
I am trying to run train.py, but I receive this error:
Traceback (most recent call last):
File "/Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/site-packages/IPython/core/interactiveshell.py", line 3505, in run_code
exec(code_obj, self.user_global_ns, self.user_ns)
File "", line 41, in
parser = Trainer.add_argparse_args(parser)
^^^^^^^^^^^^^^^^^^^^^^^^^
AttributeError: type object 'Trainer' has no attribute 'add_argparse_args'
Apparently, the pytorch lightning's Trainer does not have the method add_argparse_args in this line:
parser = Trainer.add_argparse_args(parser)
Would you please help and tell me how I can get around this error? Thank you so much :)
Best,
Arman
when computing MarginLoss, I found that the third dimension is not match at all. The raw_logits shape is torch.Size([8, 256, 64]), while the labels(after interpolate and reshape) shape is torch.Size([8, 256, 8])
and there is the error: RuntimeError: The size of tensor a (8) must match the size of tensor b (64) at non-singleton dimension 2
what's more, when choosing heart as the datasaet, the input_channels should change to 1 rather than 1,is that right?
Hi,
I have a question related to the _build_encoder function within module/mod_upcaps.py ๐
This block of code defines 6 capsule layers after the primary capsule layer:
self.encoder_output_dim = [16, 16, 8, 8, 8, self.out_channels]
self.encoder_output_atoms = [8, 8, 16, 16, 32, 64]
for i in range(len(self.encoder_output_dim)):
if i == 0:
input_dim = self.primary_caps.output_dim
input_atoms = self.primary_caps.output_atoms
else:
input_dim = self.encoder_output_dim[i - 1]
input_atoms = self.encoder_output_atoms[i - 1]
stride = 2 if i % 2 == 0 else 1
self.encoder_conv_caps.append(
ConvSlimCapsule3D(
kernel_size=self.encoder_kernel_size,
input_dim=input_dim,
output_dim=self.encoder_output_dim[i],
input_atoms=input_atoms,
output_atoms=self.encoder_output_atoms[i],
stride=stride,
padding=1,
dilation=1,
num_routing=3,
share_weight=self.share_weight,
)
)
I wanted to double check if it is correct, since in your paper you only have 2 capsule layers.
Thank you so much in advance :)
Arman
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.