Yes, it is true that there are many frameworks available for setting up datasets, such as ImageFolder
, which is commonly used in computer vision tasks. However, in the real world, the data you may encounter may not always be in a format
that can be directly used with these frameworks.
This would require you to build your own dataset and preprocess your data accordingly
With my code, you can:
- Use your dataset in your path folder
- You can customize your preprocessing steps based on the specific needs of your task
- Python
- Torch
- PIL
- Torchvision
- os
I have explain more clearly in my medium