torchvision is an R package that extends torch with tools for computer vision. It provides image loading, transformations, pre-trained model architectures, and access to commonly used datasets.
The package gives R users a consistent interface for building image-based deep learning workflows. It includes pre-trained weights for common architectures, so you can use transfer learning without training from scratch. It also provides standard image transformations and benchmark datasets, reducing boilerplate when setting up computer vision pipelines.
