Augmentation

Discussion
,
1

While training can we have control over training data. For example if I’m doing hsv_h,hsv_s,hsv_v,it will apply to all my images. For model generalisation and to avoid over fitting can I apply augmentation to only 1/4 th of data. Rest will be original

2

It applies it randomly, so it’s not applied equally strong to all images. Sometimes it’s very slight that it’s almost not noticeable.

For model generalisation and to avoid over fitting can I apply augmentation to only 1/4 th of data

Overfitting occurs more strongly when you disable augmentations, not when you enable them.

Applying HSV augmentations partially isn’t supported.