0.18.0

• Introduce the env var OPENNSFW2_KERAS=tf-keras to enforce using tf-keras if needed.

0.17.1

• Do not expose unnecessary keras_backend to top level.

0.17.0

• Add backward-compatibility support for tf-keras.

0.16.1

• Move import cv2 to the function predict_video_frames because that's the only function requires this dependency.

0.16.0

• Added preprocess_image_tensor, a tensor-based preprocessing function for use in training pipelines

0.15.2

• Remove redundant success field from HTTP API responses.

0.15.1

• Add HTTP service, see README-HTTP.md for details.

0.14.0

• Support parsing PIL.Image.Image object(s) directly other than image path(s) to predict_image and predict_images functions. Rename the original image_path/image_paths position arg to image_handle/image_handles (breaking change).
• Add Python 3.12 to CI test.

0.13.10

• Fix image overlay for GradCAM when using YAHOO preprocessing. The YAHOO preprocessing actually excludes from the model input certain pixels from each edge of the image, the GradCAM visualisation should take this into account.

0.13.9

• Fix type annotation for PIL.Image.Image.
• Use a fixed layer name global_average_pooling2d for the GlobalAveragePooling2D layer, to avoid Keras assigning a different name upon multiple model instantiations.