tf.image.sample_distorted_bounding_box

Generate a single randomly distorted bounding box for an image.

Bounding box annotations are often supplied in addition to ground-truth labels in image recognition or object localization tasks. A common technique for training such a system is to randomly distort an image while preserving its content, i.e. data augmentation. This Op outputs a randomly distorted localization of an object, i.e. bounding box, given an image_size, bounding_boxes and a series of constraints.

The output of this Op is a single bounding box that may be used to crop the original image. The output is returned as 3 tensors: begin, size and bboxes. The first 2 tensors can be fed directly into tf.slice to crop the image. The latter may be supplied to tf.image.draw_bounding_boxes to visualize what the bounding box looks like.

Bounding boxes are supplied and returned as [y_min, x_min, y_max, x_max]. The bounding box coordinates are floats in [0.0, 1.0] relative to the width and the height of the underlying image.

For example,

    # Generate a single distorted bounding box.
    begin, size, bbox_for_draw = tf.image.sample_distorted_bounding_box(
        tf.shape(image),
        bounding_boxes=bounding_boxes,
        min_object_covered=0.1)

    # Draw the bounding box in an image summary.
    image_with_box = tf.image.draw_bounding_boxes(tf.expand_dims(image, 0),
                                                  bbox_for_draw)
    tf.compat.v1.summary.image('images_with_box', image_with_box)

    # Employ the bounding box to distort the image.
    distorted_image = tf.slice(image, begin, size)

Note that if no bounding box information is available, setting use_image_if_no_bounding_boxes = true will assume there is a single implicit bounding box covering the whole image. If use_image_if_no_bounding_boxes is false and no bounding boxes are supplied, an error is raised.

For producing deterministic results given a seed value, use tf.image.stateless_sample_distorted_bounding_box. Unlike using the seed param with tf.image.random_* ops, tf.image.stateless_random_* ops guarantee the same results given the same seed independent of how many times the function is called, and independent of global seed settings (e.g. tf.random.set_seed).

image_size A Tensor. Must be one of the following types: uint8, int8, int16, int32, int64. 1-D, containing [height, width, channels].
bounding_boxes A Tensor of type float32. 3-D with shape [batch, N, 4] describing the N bounding boxes associated with the image.
seed An optional int. Defaults to 0. If seed is set to non-zero, the random number generator is seeded by the given seed. Otherwise, it is seeded by a random seed.
min_object_covered A Tensor of type float32. Defaults to 0.1. The cropped area of the image must contain at least this fraction of any bounding box supplied. The value of this parameter should be non-negative. In the case of 0, the cropped area does not need to overlap any of the bounding boxes supplied.
aspect_ratio_range An optional list of floats. Defaults to [0.75, 1.33]. The cropped area of the image must have an aspect ratio = width / height within this range.
area_range