ยป Keras API reference / KerasCV API

KerasCV API

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 standard Model.fit(), Model.evaluate(), Model.predict() flow.

Layers

Metrics

Models

Bounding box formats and utilities