KerasCV is a toolbox of modular building blocks (layers, metrics, losses, data augmentation) that computer vision engineers can leverage to quickly assemble production-grade, state-of-the-art training and inference pipelines for common use cases such as image classification, object detection, image segmentation, image data augmentation, etc.
KerasCV Layers can be used independently, or with the
keras.Model class. Layers
implement specific self contained logic such as image data augmentation, regularization
during training, and more!
KerasCV Metrics are used for model evaluation. They can be used for both
train time and post training evaluation. They can be used independently, or as part of the