Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Is the correct version of the code shared? / 是否共享了正确版本的代码? #54

Open
shahzad-ali opened this issue Oct 22, 2024 · 1 comment

Comments

@shahzad-ali
Copy link

shahzad-ali commented Oct 22, 2024

The code shared as segmamba.py is very different from the proposed SegMamba in the paper (as presented in Fig. 1). Specifically, I made the following observations immediately:

  1. The sequence of operations in the GSC module differs from that shown in Fig. 2. (a). There is no multiplication operation in its code.
  2. The Introduction section says: "To enhance the whole-volume sequential modeling of 3D features, we design a tri-orientated Mamba (ToM) module." Your response to a similar query indicates using a ready-made block from the original paper "Mamba: Linear-Time Sequence Modeling with Selective State Spaces."
  3. The paper does not acknowledge the borrowed blocks from the UNETR (i.e., UnetrBasicBlock, UnetrUpBlock, etc).
  4. The code for Feature-level Uncertainty Estimation (FUE) is missing.

The response to the above observation would be highly appreciated.

Thank you for your time and consideration.


共享为 segmamba.py 的代码与论文中提出的 SegMamba 有很大不同(如图 1 所示)。具体来说,我立即做出了以下观察:

  1. GSC 模块中的操作顺序与图 2 (a) 中所示的不同。其代码中没有乘法运算。
  2. 简介部分说:“为了增强 3D 特征的全体积顺序建模,我们设计了一个三向 Mamba (ToM) 模块。”您对 类似查询 的回复表明使用原始论文“Mamba:具有选择性状态空间的线性时间序列建模”中的现成块。
  3. 论文没有承认从 UNETR 借用的块(即 UnetrBasicBlock、UnetrUpBlock 等)。
  4. 缺少特征级不确定性估计 (FUE) 的代码。

对上述观察的回复将不胜感激。

感谢您的时间和考虑。

@ge-xing
Copy link
Owner

ge-xing commented Oct 23, 2024

Thank you for your concerns.

  1. In my experiments, I found that the addition operation is more stable than the multiplication operation in certain datasets. However, I cannot ensure this holds true for all datasets, so you might want to try both.
  2. They have installed the original Mamba block instead of the modified Mamba block available at this link. To quickly resolve their errors and run the code successfully, I suggest they remove the parameter bimamba_type == "v3" and run using the original Mamba temporally.
  3. In the acknowledgment section of this repository, I have mentioned multiple repositories, including Monai. This is because I have referred to some codes from Monai, including the blocks from UNETR you mentioned.
  4. I will correct this section as soon as possible.

Thank you for your guidance!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants