The overall layout of dataset folder root/datasets is as
├── Datasets
├── TrainingSet
├── PriorsSet
├── TestSet
In this implementation, we organized the data in the form of H5 files. The specific organization form of H5 file is
├── H5 File Name
├── Block ID1
├── data
├── label
├── Block ID2
├── data
├── label
...
├── Block IDN
├── data
├── label
Due to the huge amount of data, we split the entire data set into multiple H5 files.
The dataset folder path is root/dataset_folder.
If you want to train this on your dataset, you can rewrite the dataset.py file to adopt to your data.
Both VISoR-40 and BigNeuron dataset are public datasets. You can download them refer to VISoR-40 and BigNeuron.