You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardexpand all lines: data/README.md
+1-1
Original file line number
Diff line number
Diff line change
@@ -1,6 +1,6 @@
1
1
# Dataset Pre-processing
2
2
3
-
## General Pre-processing
3
+
## Generic Pre-processing
4
4
After downloading the dataset, please create a new folder named "images" under the folder "miniimagenet" or "cub", and put all images in this folder. The provided data loader will read images from the "images" folder by default. Of course, it is also OK to change the read path. For example, for the miniimagenet dataset, please change the line 10 of "./feat/dataloader/mini_imagenet.py" as the path of the downloaded images.
5
5
6
6
We assume all the images in the folder are the original ones (except a crop based on bounding boxes for CUB, see details below), and the data loader will do transformations on those raw images, such as resize and normalization. All the images will be resized as 84x84 for ConNet backbone, and 80x80 for ResNet backbone.
0 commit comments