error in docs

[jeremy-rutman]

The docs here for bounding boxes have an error :
The paragraph reading

For example, if an image is 100 x 200 pixels and the bounding box is [0.1, 0.2, 0.5, 0.9], the bottom-left and upper-right coordinates of the bounding box will be (10, 40) to (50, 180).

would better go something like:

For example, if an image is 100 x 200 pixels and the bounding box is [0.1, 0.2, 0.5, 0.9], the upper-left and bottom-right coordinates of the bounding box will be (10, 40) to (50, 180), using a somewhat idiosyncratic (y,x) notation, or (40,10) to (180,50) using the rather more common (x,y) notation for points in a plane. All this is with 'usual' coordinate axes (origin in the top left corner, increasing x to right and increasing y going down)

[clemens-tolboom]

Is it really height x width ? Yes as said on https://www.tensorflow.org/versions/r0.12/api_docs/python/image/working_with_bounding_boxes

The coordinates of the each bounding box in boxes are encoded as [y_min, x_min, y_max, x_max].

You could create a PR for around line 700 in
https://github.com/tensorflow/tensorflow/blob/master/tensorflow/core/ops/image_ops.cc#L700

I agree It is not intuitive but it's mentioned as swapped.

---

Wouldn't it be better to mention width and height to trigger the users?

For example, if an image is 100 height x 200 width pixels

or

For example, if an image is 100 x 200 (height x width) pixels

[jeremy-rutman]

I tried doing a pr 11889 as you suggested, thanks.

[aselle]

see #11889 for fix.

[divyam3897]

Hi, it should be closed if fixed?

[jeremy-rutman]

the pr was merged so I suppse you are right, will close if I can