不同分辨率patch 特征如何获取？

[CV-ZhangXin]

非常感谢你们的杰出工作，

我想问一下，下载TCGA-RCC数据集之后，怎么获得低分辨和高分辨率下的Patch特征。

[CV-ZhangXin]

@Jiangbo-Shi

[smart0eddie]

I modify a few arguments as the following.
However, I'm not sure if the way I process the patches of different scale is correct.
The reproduction results I got is about 5% lower than their paper.

create_patches_fp.py
// 10x
// "--patch_size", "1024",
// "--step_size", "1024",

// 5x
"--patch_size", "2048",
"--step_size", "2048",

feature_extraction/patch_generation.py
//"--SAVE_FOLDER_PATH", "$PATH_TO_1024_PATCH"
//"--define_patch_size", "1024"

"--SAVE_FOLDER_PATH", "$PATH_TO_2048_PATCH"
"--define_patch_size", "2048"

feature_extraction/patch_extraction.py
//"--patches_path", "$PATH_TO_1024_PATCH"
//"--library_path", "$PATH_TO_1024_FEATURES"

"--patches_path", "$PATH_TO_2048_PATCH"
"--library_path", "$PATH_TO_2048_FEATURES"

main.py
"--data_folder_s", "$PATH_TO_2048_FEATURES",
"--data_folder_l", "$PATH_TO_1024_FEATURES",

[CV-ZhangXin]

非常感谢你，祝你新年快乐！

…

---原始邮件--- 发件人: ***@***.***> 发送时间: 2025年1月29日(周三) 凌晨3:24 收件人: ***@***.***>; 抄送: "Zhang ***@***.******@***.***>; 主题: Re: [Jiangbo-Shi/ViLa-MIL] 不同分辨率patch 特征如何获取？ (Issue #16) I modify a few arguments as the following. However, I'm not sure if the way I process the patches of different scale is correct. The reproduction results I got is about 5% lower than their paper. create_patches_fp.py // 10x // "--patch_size", "1024", // "--step_size", "1024", // 5x "--patch_size", "2048", "--step_size", "2048", feature_extraction/patch_generation.py //"--SAVE_FOLDER_PATH", "$PATH_TO_1024_PATCH" //"--define_patch_size", "1024" "--SAVE_FOLDER_PATH", "$PATH_TO_2048_PATCH" "--define_patch_size", "2048" feature_extraction/patch_extraction.py //"--patches_path", "$PATH_TO_1024_PATCH" //"--library_path", "$PATH_TO_1024_FEATURES" "--patches_path", "$PATH_TO_2048_PATCH" "--library_path", "$PATH_TO_2048_FEATURES" main.py "--data_folder_s", "$PATH_TO_2048_FEATURES", "--data_folder_l", "$PATH_TO_1024_FEATURES", — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: ***@***.***>