Image
Built-in pipelines for preprocessing, agumentating and predicting.
AugmentImage
paz.pipelines.image.AugmentImage()
Augments an RGB image by randomly changing contrast, brightness saturation and hue.
PreprocessImage
paz.pipelines.image.PreprocessImage(shape, mean=(104, 117, 123))
Preprocess RGB image by resizing it to the given shape
. If a
mean
is given it is substracted from image and it not the image gets
normalized.
Arguments
- shape: List of two Ints.
- mean: List of three Ints indicating the per-channel mean to be subtracted.
DecoderPredictor
paz.pipelines.image.DecoderPredictor(decoder)
Pipeline for predicting decoded image from a latent vector.
Arguments
- model: Keras model.
EncoderPredictor
paz.pipelines.image.EncoderPredictor(encoder)
Pipeline for predicting latent vector of an encoder.
Arguments
- model: Keras model.
PreprocessImageHigherHRNet
paz.pipelines.image.PreprocessImageHigherHRNet(scaling_factor=200, input_size=512, multiple=64)
Transform the image according to the HigherHRNet model requirement. Arguments
- scaling_factor: Int. scale factor for image dimensions.
- input_size: Int. resize the first dimension of image to input size.
- inverse: Boolean. Reverse the affine transform input.
- image: Numpy array. Input image
Returns
- image: resized and transformed image
- center: center of the image
- scale: scaled image dimensions