Wrapper for inserting float features into Example proto.

Returns `True` if `x` is a symbolic tensor-like object. Args: x: A python object to check. Returns: `True` if `x` is a `tf.Tensor`

Wrapper for inserting bytes features into Example proto.

Wrapper for inserting int64 features into Example proto.

A polymorphic assert, works with tensors and boolean expressions. If `cond` is not a tensor, behave like an ordinary assert statement, except

Build an Example proto for an image example. Args: image_data: string, JPEG encoding of RGB image; labels: list of integers, identifie

Tell whether the 2D coordinate (x, y) is valid or not. If valid, it should be on an h x w image

Returns the dimensions of an image tensor. Args: image: A 3-D Tensor of shape `[height, width, channels]`. Returns: A list of `[he

Creates a new local variable. Args: name: The name of the new or existing variable. shape: Shape of the new or existing variable.

Assert that we are working with properly shaped image. Args: image: 3-D Tensor of shape [height, width, channels] require_static: If

generate anchors on on layer return a ndarray with shape (h_l, w_l, 4), and the last dimmension in the order:[cx, cy, w, h]

calculate the height ratio between anchor and rect

Adapt bounding boxes to crop or pad operations. Coordinates are always supposed to be relative to the image. Arguments: bboxes: Tensor

Compute relative intersection between a reference box and a collection of bounding boxes. Namely, compute the quotient between intersection ar

Step 2 to 4

Args: xs: numpy ndarray with shape=(N,4). N is the number of oriented bboxes. 4 contains [x1, x2, x3, x4] ys: numpy ndarray with

Subtracts the given means from each image channel. Returns: the centered image.

Add image with bounding boxes to summary.

Transform the rects from opencv method minAreaRect to our rects. Step 1 of Figure 5 in seglink paper In cv2.minAreaRect, the w, h and theta

test _generate_anchors_one_layer method by visualizing it in an image.

Computes func(x, sel), with sel sampled from [0...num_cases-1]. Args: x: input Tensor. func: Python function to apply. nu

Convert Seglink bboxes to xys, i.e., eight points The `image_shape` is used to to make sure all points return are valid, i.e., within image area

Step 3 in Figure 5 in seglink paper The crop operation is operated only on the x direction. Args: bbox: a horizontal bbox with shape =

Distort the color of a Tensor image. Each color distortion is non-commutative and thus ordering of the color ops matters. Ideally we would ra

Generates cropped_image using a one of the bboxes randomly distorted. See `tf.image.sample_distorted_bounding_box` for more documentation. A

Args: seg_locs: a ndarray with shape = (N, 5). It contains the abolute values of segment locations Return: seg_offsets, i.e.

Set the shape to 3 dimensional if we don't know anything else. Args: image: original image size result: flipped or transformed image

Flip bounding boxes coordinates.

group segments based on their scores and links. Return: segment groups as a list, consisting of list of segment indexes, reprensting a group

Match anchors to text boxes. Return: seg_labels: shape = (N,), the seg_labels of segments. each value is the index of matched box i

Preprocess an image for evaluation. Args: image: A `Tensor` representing an image of arbitrary size. out_shape: Output shape afte

Preprocesses the given image for training. Note that the actual resizing scale is sampled from [`resize_size_min`, `resize_size_max`].