Your Web News in One Place

Help Webnuz

Referal links:

Sign up for GreenGeeks web hosting
September 14, 2021 06:38 pm GMT

A Visualizer for PyTorch Image Transformations

Alt Text

Alt Text

Tldr; Check out the app here.

Image augmentation is a common technique used when training computer vision models in order to generate artificial training data by transforming in your actual training data, for example, random rotations and shifts.

Alt TextWe created 4 new images of cats!

However these augmentations can often be a source of subtle bugs. For example here is an typical PyTorch transform pipeline:

from torchvision import transformstransform = transforms.Compose([    transforms.RandomAffine(degrees=360, translate=(0.64, 0.98), scale=(0.81, 2.85), shear=(0.1, 0.5), fill=0, interpolation=InterpolationMode.NEAREST),    transforms.ColorJitter(brightness=0.62, contrast=0.3, saturation=0.44, hue=0.24),    transforms.RandomVerticalFlip(p=0.45)])

Do you see anything obviously wrong with it?

Well, lets see what happens when we actually try to apply these transformations to our cat.

Alt TextHalf of these dont even look like cats!

As you can see the transformations were not properly tuned and resulted in a significant number of images being completely unrecognizable. If we train the model on this augmented dataset, its no longer learning what a cat looks like.
These sorts of bugs are tricky because no errors will be raised. Instead the result would be that the model will not perform as well on the un-augmented test dataset as it could have.

Introducing a PyTorch transforms visualizer

Alt Text

You can use this tool to develop and sanity check your transforms on actual images before using them in a training script. It supports all the transforms provided in the torchvision.transforms package.

Check it out here!


Original Link: https://dev.to/mage_ai/a-visualizer-for-pytorch-image-transformations-5dfg

Share this article:    Share on Facebook
View Full Article

Dev To

An online community for sharing and discovering great ideas, having debates, and making friends

More About this Source Visit Dev To