TorchVision Transforms: Image Preprocessing in PyTorch

Ben Cook • Posted 2021-03-19 • Last updated 2021-10-21

TorchVision, a PyTorch computer vision package, has a great API for image pre-processing in its torchvision.transforms module. This post gives some basic usage examples, describes the API and shows you how to create and use custom image transforms.

Binary Cross Entropy Explained

Ben Cook • Posted 2021-02-22 • Last updated 2021-10-21

A simple NumPy implementation of the binary cross entropy loss function and some intuition about why it works.

NumPy Meshgrid: Understanding np.meshgrid()

Ben Cook • Posted 2021-02-09 • Last updated 2021-10-21
numpy meshgrid loss surface

You can create multi-dimensional coordinate arrays using the np.meshgrid() function, which is also available in PyTorch and TensorFlow. But watch out! PyTorch uses different indexing by default so the results might not be the same.

SageMaker Studio Quick Start

Ben Cook • Posted 2021-02-08 • Last updated 2021-10-21
sagemaker studio launcher

A step-by-step quick start guide for SageMaker Studio. Start a Studio session, launch a notebook on a GPU instance and run object detection inference with a detectron2 pre-trained model.

PyTorch One Hot Encoding

Ben Cook • Posted 2021-02-02 • Last updated 2021-12-13

PyTorch has a one_hot() function for converting class indices to one-hot encoded targets.