Computer Vision Apps

torchapp includes subclasses to assist in building apps for computer vision tasks.

Reference

torchapp.vision.TorchvisionModelEnum: alias of torchapp.vision.TorchvisionModelName

class torchapp.vision.UNetApp

A TorchApp which uses a base model from torchvision which is modified.

Useful for image segmentation, super-resolution or colorization. The default base torchvision model is resnet18.

For more information see: Olaf Ronneberger, Philipp Fischer, Thomas Brox, U-Net: Convolutional Networks for Biomedical Image Segmentation, https://arxiv.org/abs/1505.04597 https://github.com/fastai/fastbook/blob/master/15_arch_details.ipynb

build_learner_func()

Returns unet_learner

class torchapp.vision.VisionApp

A TorchApp which uses a model from torchvision.

The default base torchvision model is resnet18.

torchapp.vision.torchvision_model_choices() → List[str]

Returns a list of function names in torchvision.models which can produce torch modules.