pesser / edflow

Reduce boilerplate code for your ML projects. TensorFlow and PyTorch.
https://edflow.readthedocs.io/
MIT License
25 stars 13 forks source link

How to automatically resize input images in template_tfe example #158

Closed theRealSuperMario closed 4 years ago

theRealSuperMario commented 5 years ago

Is your feature request related to a problem? Please describe. In the old Iterators before tfe, we could simply use the spatial_size key from the config to resize the image. How is that possible in the new iterator?

Describe the solution you'd like I would like to keep the option from the config to simple resize from the config file

Describe alternatives you've considered

pesser commented 5 years ago

this is not an issue of the new iterator. spatial_size is used by quite a few datasets to resize images appropriately. if it doesn't work, it means the dataset does not care to resize. Quick fix is to add the resizing support to the dataset of interest but maybe we want to move towards more configurable dataset preprocessing chains? Something like

preprocessing:
    chain:
        - Resize
        - Crop
    Resize:
        spatial_size: 256
    Crop:
        size: 128
pesser commented 4 years ago

closing as this is completely dataset dependent. Please create a new issue if you want to develop a generic solution to preprocessing chains