数据集下载不了

[laiqirong]

请问这个数据集是原始心电数据吗

[onlyzdd]

你可以从 CPSC2018、Physionet 2020 Challenge 或者这里下载数据集。Dropbox 上的文件是原始数据，只是组织上更规整一些。

[rdyan0053]

Hello, I have downloaded the CPSC2018 dataset. But I found that the dataset and the code preprocess.py do not match.

For example, in the CPSC2018 dataset, here are 3 folders TrainingSet1, TrainingSet2, TrainingSet3. And in these folders, here only mat files, no hea files. But in the preprocess.py, the line recordpaths = glob(os.path.join(data_dir, '*.hea')) is to find all paths of hea file.

So, I have a little confusion.

[rdyan0053]

sorry, i have see the dataset in dropbox, thank you

[onlyzdd]

Hello, I have downloaded the CPSC2018 dataset. But I found that the dataset and the code preprocess.py do not match.

For example, in the CPSC2018 dataset, here are 3 folders TrainingSet1, TrainingSet2, TrainingSet3. And in these folders, here only mat files, no hea files. But in the preprocess.py, the line recordpaths = glob(os.path.join(data_dir, '*.hea')) is to find all paths of hea file.

So, I have a little confusion.

@rdyan0053 They are exactly the same data but in different formats. New format of the CPSC2018 data is avaliable as part of training data in the PhysioNet 2020 Challange. It's better orangized and can be easily loaded with the wfdb package.

If you're not using the Dropbox one, you may need to make some changes on data loading and preprocessing.

[rdyan0053]

Hello, I have downloaded the CPSC2018 dataset. But I found that the dataset and the code preprocess.py do not match. For example, in the CPSC2018 dataset, here are 3 folders TrainingSet1, TrainingSet2, TrainingSet3. And in these folders, here only mat files, no hea files. But in the preprocess.py, the line recordpaths = glob(os.path.join(data_dir, '*.hea')) is to find all paths of hea file. So, I have a little confusion.

@rdyan0053 They are exactly the same data but in different formats. New format of the CPSC2018 data is avaliable as part of training data in the PhysioNet 2020 Challange. It's better orangized and can be easily loaded with the wfdb package.

If you're not using the Dropbox one, you may need to make some changes on data loading and preprocessing.

Right. Thank you very much!!!