View Source Bumblebee.Vision.BitFeaturizer (Bumblebee v0.5.3)
BiT featurizer for image data.
Configuration
:resize- whether to resize the input to the given:size. Defaults totrue:size- the size to resize the input to, either%{height: ..., width: ...}or%{shortest_edge: ...}. Only has an effect if:resizeistrue. Defaults to%{shortest_edge: 448}:resize_method- the resizing method, either of:nearest,:bilinear,:bicubic,:lanczos3,:lanczos5. Defaults to:bicubic:center_crop- whether to crop the input at the center. If the input size is smaller than:crop_sizealong any edge, the image is padded with zeros and then center cropped . Defaults totrue:crop_size- the size to center crop the image to, given as%{height: ..., width: ...}. Only has an effect if:center_cropistrue. Defaults to%{width: 448, height: 448}:rescale- whether to rescale the input by the given:rescale_factor. Defaults totrue:rescale_factor- the factor by which to rescale the input. A single number Only has an effect if:rescaleistrue. Defaults to0.00392156862745098:normalize- whether or not to normalize the input with mean and standard deviation. Defaults totrue:image_mean- the sequence of mean values for each channel, to be used when normalizing images. Defaults to[0.5, 0.5, 0.5]:image_std- the sequence of standard deviations for each channel, to be used when normalizing images. Defaults to[0.5, 0.5, 0.5]