Where is the required txt and folder?

[jk456445]

I used the preprocessing code you provided to preprocess the downloaded human3d_egobody_pcd_data.zip (https://egobody.ethz.ch/data/dataset/human3d_egobody_pcd_data.zip)files, but it says it can't find the required txt file(train_list.txt and validation_list.txt). I also didn't see the specified file in it; instead, there are other txt files. Could you please confirm if the required file according to the code is included among them?

Additionally, the code requests to find items under the folder data/raw/egobody/scenes/.

However, I don't have this folder. Under the egobody folder, there are various recording folders. Could you please confirm whether there is an issue with the raw data I downloaded or if it is a problem within the code?

[aycatakmaz]

Hi,

Thank you very much for your interest in our work, and sorry for the late reply!

The file names train_list.txt and validation_list.txt are from a previous iteration of the dataset, you can simply rename the split_train.txt and split_val.txt. The files you already downloaded should be essentially the same, I think we simply renamed them at some point.

Regarding the second issue, I checked and I can confirm that the requested file recording_20210907_S02_S01_01/main/frame_01581.ply exists in the raw dataset human3d_egobody_pcd_data.zip. I think the problem is that in the error log you sent, the path is shown as data/raw/egobody/scenes/recording_20210907_S02_S01_01/main/frame_01581.ply, however according to your folder structure this file is located at data/raw/egobody/recording_20210907_S02_S01_01/main/frame_01581.ply. So there is an additional scenes subfolder in the path that the code looks for, due to this line in the code, please see lines L223-238.

Thank you very much for noticing the mismatch between the downloaded raw files and the preprocessing code! We will try to push an update to fix it as soon as we can, but in the meantime, I think you can simply replace the part between L223-238 in datasets/preprocessing/humanseg_preprocessing.py with the following:

  for mode in self.modes:
      with open(f"{data_dir}/{mode}_list.txt") as file:
          if self.dataset == "egobody" and mode == "validation":
              self.files[mode] = natsorted(
                  [
                      f"{self.data_dir}/{line.strip()}"
                      for line in file
                  ]
              )
          else:
              self.files[mode] = natsorted(
                  [
                      f"{self.data_dir}/{line.strip()}"
                      for line in file
                  ]
              )

@JonasSchult could you please verify that my proposed change would fix the issue? If so, I can make a commit!

@jk456445 Let us know if this solves the problem!

Best,
Ayca

[Aeyu17]

Hello!

I noticed that this issue never got resolved while I had the same issue, and this solution worked for me. Thank you so much!

[seakafuka]

Hi, 你好，

Thank you very much for your interest in our work, and sorry for the late reply!非常感谢您对我们工作的关注，很抱歉回复晚了！

The file names train_list.txt and validation_list.txt are from a previous iteration of the dataset, you can simply rename the split_train.txt and split_val.txt. The files you already downloaded should be essentially the same, I think we simply renamed them at some point.文件名train_list.txt和validation_list.txt来自数据集的先前迭代，您可以简单地重命名split_train.txt和split_val.txt 。您已经下载的文件应该本质上是相同的，我认为我们只是在某个时候重命名了它们。

Regarding the second issue, I checked and I can confirm that the requested file recording_20210907_S02_S01_01/main/frame_01581.ply exists in the raw dataset human3d_egobody_pcd_data.zip. I think the problem is that in the error log you sent, the path is shown as data/raw/egobody/scenes/recording_20210907_S02_S01_01/main/frame_01581.ply, however according to your folder structure this file is located at data/raw/egobody/recording_20210907_S02_S01_01/main/frame_01581.ply. So there is an additional scenes subfolder in the path that the code looks for, due to this line in the code, please see lines L223-238.关于第二个问题，我检查并可以确认原始数据集human3d_egobody_pcd_data.zip中存在所请求的文件recording_20210907_S02_S01_01/main/frame_01581.ply 。我认为问题是在您发送的错误日志中，路径显示为data/raw/egobody/scenes/recording_20210907_S02_S01_01/main/frame_01581.ply ，但是根据您的文件夹结构，该文件位于data/raw/egobody/recording_20210907_S02_S01_01/main/frame_01581.ply 。因此，代码查找的路径中存在一个额外的scenes子文件夹，由于代码中的这一行，请参见第 L223-238 行。

Thank you very much for noticing the mismatch between the downloaded raw files and the preprocessing code! We will try to push an update to fix it as soon as we can, but in the meantime, I think you can simply replace the part between L223-238 in datasets/preprocessing/humanseg_preprocessing.py with the following:非常感谢您注意到下载的原始文件和预处理代码之间的不匹配！我们将尝试尽快推送更新来修复它，但与此同时，我认为您可以简单地将datasets/preprocessing/ humanseg_preprocessing.py中 L223-238 之间的部分替换为以下内容：

  for mode in self.modes:
      with open(f"{data_dir}/{mode}_list.txt") as file:
          if self.dataset == "egobody" and mode == "validation":
              self.files[mode] = natsorted(
                  [
                      f"{self.data_dir}/{line.strip()}"
                      for line in file
                  ]
              )
          else:
              self.files[mode] = natsorted(
                  [
                      f"{self.data_dir}/{line.strip()}"
                      for line in file
                  ]
              )

@JonasSchult could you please verify that my proposed change would fix the issue? If so, I can make a commit!您能否确认我提出的更改是否可以解决该问题？如果是这样，我可以做出承诺！

@jk456445 Let us know if this solves the problem!让我们知道这是否可以解决问题！

Best, 最好的， Ayca 艾卡

Hi，where is the txt of Synthetic Data? not in your result.zip

[wyedoubleyou]

Hi, can I follow up on the txt of synthetic data, where can I obtain the train_list.txt and validation_list.txt for Synthetic Data?