Module: tfm.vision.preprocess_ops

Preprocessing ops.

Classes

class range: range(stop) -> range object range(start, stop[, step]) -> range object

Functions

center_crop_image(...): Center crop a square shape slice from the input image.

center_crop_image_v2(...): Center crop a square shape slice from the input image.

clip_or_pad_to_fixed_size(...): Pads data to a fixed length at the first dimension.

color_jitter(...): Applies color jitter to an image, similarly to torchvision`s ColorJitter.

compute_padded_size(...): Compute the padded size given the desired size and the stride.

horizontal_flip_boxes(...): Flips normalized boxes horizontally.

horizontal_flip_image(...): Flips image horizontally.

horizontal_flip_masks(...): Flips masks horizontally.

normalize_image(...): Normalizes the image to zero mean and unit variance.

normalize_scaled_float_image(...): Normalizes a scaled float image to zero mean and unit variance.

random_brightness(...): Jitters brightness of an image.

random_contrast(...): Jitters contrast of an image, similarly to torchvision`s ColorJitter.

random_crop(...): Randomly crop the image and boxes, filtering labels.

random_crop_image(...): Randomly crop an arbitrary shaped slice from the input image.

random_crop_image_v2(...): Randomly crop an arbitrary shaped slice from the input image.

random_crop_image_with_boxes_and_labels(...): Crops a random slice from the input image.

random_horizontal_flip(...): Randomly flips input image and bounding boxes horizontally.

random_horizontal_flip_with_roi(...): Randomly flips input image and bounding boxes horizontally.

random_saturation(...): Jitters saturation of an image, similarly to torchvision`s ColorJitter.

random_vertical_flip(...): Randomly flips input image and bounding boxes vertically.

resize_and_crop_boxes(...): Resizes boxes to output size with scale and offset.

resize_and_crop_image(...): Resizes the input image to output size (RetinaNet style).

resize_and_crop_image_v2(...): Resizes the input image to output size (Faster R-CNN style).

resize_and_crop_masks(...): Resizes boxes to output size with scale and offset.

resize_image(...): Resize image with size and max_size.

vertical_flip_boxes(...): Flips normalized boxes vertically.

CENTER_CROP_FRACTION 0.875
MEAN_NORM (0.485, 0.456, 0.406)
MEAN_RGB (123.675, 116.28, 103.53)
MEDIAN_RGB (128.0, 128.0, 128.0)
STDDEV_NORM (0.229, 0.224, 0.225)
STDDEV_RGB (58.395, 57.120000000000005, 57.375)