search

mihaidusmanu / d2-net

D2-Net: A Trainable CNN for Joint Description and Detection of Local Features
Other
761 stars 163 forks source link

Inquiry about D2Net for Brain Tumor Segmentation on BraTS2018 Dataset #99

Closed MaggieLSY closed 9 months ago

MaggieLSY commented 1 year ago

Hi, I am currently working on applying missing modality models for brain tumor segmentation, and I have some questions regarding your experiments on the BraTS2018 dataset. I hope you don't mind sharing some details with me.

  1. Having carefully studied the U2NET paper, I found the section on the three-fold data partitioning particularly intriguing. To further explore this aspect, I kindly request if you could provide me with the list that outlines the specific division of data into three folds. This information would greatly enhance my understanding of the experimental setup and facilitate my own research.

  2. While reading the paper, I noticed that the authors mentioned the use of three-fold cross-validation. However, the reported results consist of a single value rather than the more conventional form of mean and standard deviation. I am curious to know how the performance metric was calculated and whether it represents an average value across the three folds.

Your valuable insights into these inquiries would be immensely helpful to me. I greatly appreciate your time and effort in addressing my questions. Thank you in advance for your kind assistance.

Looking forward to your response.

mihaidusmanu commented 1 year ago

It sounds like you are confusing D2-Net (this repository) with U2NET (not sure what that is).

D2-Net is not used for Brain Tumour Segmentation but for local feature matching.

On Sat, 24 Jun 2023, 18:05 MaggieLSY, @.***> wrote:

Hi, I am currently working on applying missing modality models for brain tumor segmentation, and I have some questions regarding your experiments on the BraTS2018 dataset. I hope you don't mind sharing some details with me.

1.

Having carefully studied the U2NET paper, I found the section on the three-fold data partitioning particularly intriguing. To further explore this aspect, I kindly request if you could provide me with the list that outlines the specific division of data into three folds. This information would greatly enhance my understanding of the experimental setup and facilitate my own research. 2.

While reading the paper, I noticed that the authors mentioned the use of three-fold cross-validation. However, the reported results consist of a single value rather than the more conventional form of mean and standard deviation. I am curious to know how the performance metric was calculated and whether it represents an average value across the three folds.

Your valuable insights into these inquiries would be immensely helpful to me. I greatly appreciate your time and effort in addressing my questions. Thank you in advance for your kind assistance.

Looking forward to your response.

— Reply to this email directly, view it on GitHub https://github.com/mihaidusmanu/d2-net/issues/99, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABXQNQH6YMRCJFTZCJHJUDDXM4F6JANCNFSM6AAAAAAZSSX3UU . You are receiving this because you are subscribed to this thread.Message ID: @.***>