Part-1: Dataloaders for different scenarios of data augmentation in PyTorch

This video and upcoming videos will show how we can create data loaders for different scenarios of data augmentation in PyTorch.
Scenario -1: one augmented image from one input image.
Scenario-2: Using augmentation for increasing dataset, multiple augmented images from one input image.
Scenario-3: Using augmentation for increasing dataset of selective classes.
Links:
Codes: github.com/anu...
Different Augmentations using transformations: pytorch.org/vi...
Dogs-vs-cats dataset: www.kaggle.com...
#DeepLearning #PyTorch #Dataloaders in Pytorch #Data Augmentation in Pytorch #DataAugmentation #Augmentation for increasing dataset size
#Custom Collate Function in PyTorch

[@maxinteltech3321]

Great video. Thanks

[@anujshah645]

Glad it was informative

[@Superrx78eddy]

Great stuff!!! Loving it. Is there any video in reducing the over fitting or improve the training performance on this topic?

[@leo.y.comprendo]

Thanks for the video! I have a question, when doing the training loop, should I use a train data loader with no augmentation and another one with augmentation to double the data available? thnks

[@anujshah645]

Well usually augmentation is used for training data alone, you can double the data by using augmentation on the data in training set either doubling all the sample or doubling samples of any selective class