Is it meaningful/allowed doing augmentation on the training set before the creation of the model, in order to achieve the same number of healthy/unhealthy plants in a new training set used in model.fit()?
Posted by: Penzo @ Nov. 11, 2023, 8:31 p.m.More in general, can we manipulate the dataset not only by removing the photos at will (seeing as this has already been done for some of them), but actually by adding "unique" ones starting from those available, doing the augmentation only on a portion (e.g. and/or healthy and/or unhealthy plants) staying? No external photos would be used doing so
Posted by: Penzo @ Nov. 14, 2023, 8:10 a.m.