for a given Image Object i want to run it through list of tfms i have for say 4 times so that i can do average of predictions made for each run of tfms.
ds[0][0].apply_tfms(tfms)
-
how i can do that ?
-
if i do some thing like this
Can i assume that apply_tfms will have applied list of all transforms with probabilities define for randomness ?l =[learn2.predict(learn2.data.train_ds[0][0].apply_tfms(learn2.data.train_ds.tfms ))[2] for _ in range(4)] torch.stack(l ).size()
output 4*6