Collection of useful and practical scripts used for Computer Vision
-
Presplit Image Classification Dataset to train and test split.
Suitable for DL4J's Image DatasetIterator, Pytorch's ImageFolder, tf.keras's ImageGenerator and fastai ImageDataBunch.from_folder.
Do not rely on sklearn or DL framework's splitter and shuffler. -
Presplit Object Detection dataset
Supports Pascal VOC, randomly samples and copies the images and annotation files(.xml).
Modify accordingly for your own scenario