Load loads an image from a file.

LoadAndResize loads and resizes an image, preserve the aspect ratio by taking a center crop.

LoadDir loads all the images in a directory.

LoadMNISTDir loads all MNIST data from a given directory to Dataset.

Pad the image using reflections and take some random crops.

Applies cutout: randomly remove some square areas in the original images.

RandomFlip randomly applies horizontal flips This expects a 4 dimension NCHW tensor and returns a tensor with an identical shape.

Resize resizes an image.

ResizePreserveAspectRatio resizes an image, preserve the aspect ratio by taking a center crop.

ResNet101 creates a ResNet-101 model.

ResNet101 creates a ResNet-101 model without final fully connfected layer.

ResNet150 creates a ResNet-150 model without final fully connfected layer.

ResNet152 creates a ResNet-152 model.

ResNet18 creates a ResNet-18 model.

ResNet18 creates a ResNet-18 model without final fully connfected layer.

ResNet34 creates a ResNet-34 model.

ResNet34 creates a ResNet-34 model without final fully connfected layer.

ResNet50 creates a ResNet-50 model.

ResNet50 creates a ResNet-50 model without final fully connfected layer.

Save saves an image to a file.