You will have one new version of it per epoch, each time you iterate through your data.
4 Likes
Why do we have both batch_tfms are also applied on images arenāt they
How is the RandomResizedCrop be applied to validation data?
A post was merged into an existing topic: Lesson 3 - Non-beginner discussion
Based on my understanding, in each epoch, the image will be cropped to a small area of the image and zoomed in. I may be wrong though.
1 Like
Itās a center crop.
1 Like
The min_scale takes 30% of the image at the timeā¦ is there an industry percentage that is recommended to use here?
How can we add diffferent augmentations for Train and Validation set?
4 Likes
My understanding is:
batch_tfms applies the SAME transform to all images in your batch.
item_tfms applies different transforms to all images in your batch.
This is effectively what itās doing. The model sees a lot more variations of the input images than if you didnāt have this randomized transform.
No, it takes a random percentage between 30% and 100% of the area (usually we use 0.35).
2 Likes
what is the different between mult=2 or 1 or 3 etc?
1 Like
Sorry for posting in the wrong forum, @sgugger!
The book states ā¦
"On each epoch (which is one complete pass through all of our images in the dataset) we randomly select a different part of each image. "
But I thought the transformations were applied per batch ā¦ not per epoch.
1 Like
Think any of these transforms be better than padding (adding empty useless pixels).
That is incorrect, batch_tfms can apply a different transformation to each image. Rotation for instance, or flip, are random per image.
1 Like
A post was merged into an existing topic: Lesson 3 - Non-beginner discussion
On the validation set, we center crop the image if itās ratio isnāt in the range (to the minmum or maximum value) then resize. Link is:
Documentation of RandomResizedCrop
2 Likes
At each epoch you get a new transformation.
Are there pre-packaged transformations that are not affine? Things like lightness, hue, sharpnessā¦