from torch.utils.tensorboard import SummaryWriter import torchvision # Initialize SummaryWriter writer = SummaryWriter() # Load images from a directory image_dir = './images' images = torchvision.datasets.ImageFolder(image_dir) # Add images to TensorBoard for i, (image, label) in enumerate(images): writer.add_image('Image {}'.format(i), image) # Close SummaryWriter writer.close()
from torch.utils.tensorboard import SummaryWriter import torch # Initialize SummaryWriter writer = SummaryWriter() # Create a PyTorch tensor images_tensor = torch.randn(10, 3, 32, 32) # Add images to TensorBoard writer.add_images('Images', images_tensor) # Close SummaryWriter writer.close()In both examples, we use SummaryWriter to initialize the TensorBoard server and add images to it. The first example demonstrates how to add images from a directory using the ImageFolder class from torchvision. The second example shows how to add images from a PyTorch tensor. Overall, SummaryWriter and the torch.utils.tensorboard package library are powerful tools for visualizing and analyzing data in neural network training. They can be used to gain insights into the model's performance, identify potential issues, and optimize the training process.