Hello all. In the task I am working on, I am incorrectly reading the volume images with CacheDataset, because the shape they have is HxWxD. When creating the loader, I want shape (B,C,D,H,W), and I am getting (B,C,H,W,D). I saw the transformation transpose, that should be enough passing the correct indexes as arguments. Nevertheless, I am dealing with a dictionary in my loaders, with keys ['image'] and ['label']. Would it be possible to have a "Transposed" in the docs?
Thank you for your time
Hello all. In the task I am working on, I am incorrectly reading the volume images with CacheDataset, because the shape they have is HxWxD. When creating the loader, I want shape (B,C,D,H,W), and I am getting (B,C,H,W,D). I saw the transformation transpose, that should be enough passing the correct indexes as arguments. Nevertheless, I am dealing with a dictionary in my loaders, with keys ['image'] and ['label']. Would it be possible to have a "Transposed" in the docs?
Thank you for your time