How can I train and test on a dataset with training, validation and testing set in file ?

[dangmanhtruong1995]

Hi, I am trying to train a dataset for anomaly detection. My data file structure is basically same as with Keras's ImageDataGenerator: One train, val and test set, each have folders of each class (here is normal and abnormal). However I have noticed that in the example on the main page in github you had this:

dataset:
name:
format: folder
path: <path/to/folder/dataset>
normal_dir: normal # name of the folder containing normal images.
abnormal_dir: abnormal # name of the folder containing abnormal images.
normal_test_dir: null # name of the folder containing normal test images.
task: segmentation # classification or segmentation
mask: <path/to/mask/annotations> #optional
extensions: null
split_ratio: 0.2 # ratio of the normal images that will be used to create a test split
image_size: 256
train_batch_size: 32
test_batch_size: 32
num_workers: 8
transform_config:
train: null
val: null
create_validation_set: true
tiling:
apply: false
tile_size: null
stride: null
remove_border_count: 0
use_random_tiling: False
random_tile_count: 16

I have noticed that there is only normal_test_dir and not abnormal_test_dir. May I ask how can I do this then ? Thank you very much.

[samet-akcay]

@dangmanhtruong1995, it is because anomaly tasks do not need abnormal images during training, but only normal images. In this case, any abnormal images will be part of the test directory.

One of the limitations of the Folder dataset format is that it is missing validation path support. It always creates the validation set by splitting the test set. We could perhaps consider adding option so you could use your own validation set.