resize the bbox with true size

[ttaoREtw]

Great work!

Here is why I create this PR:

(1) Original script uses
image_size = (images.image_sizes[idx][1], images.image_sizes[idx][0])
to resize bbox, which is incorrect because this size is padded with zeros (padding is done in def to_image_list(tensors, size_divisible)).
We should use original scaled size to resize the bbox, and in my experiment, this improves the performance a lot.

(2) Also, in the data/README, you say that the format is "x, y, w, h".
However, you use mode=xyxy in your implementation (BoxList has default mode='xyxy').
So, I think you should slightly modify your README.

Thank you, and Vilbert is a very good idea and project!

[facebook-github-bot]

Hi @ttaoREtw!

Thank you for your pull request and welcome to our community. We require contributors to sign our Contributor License Agreement, and we don't seem to have you on file.

In order for us to review and merge your code, please sign at https://code.facebook.com/cla. If you are contributing on behalf of someone else (eg your employer), the individual CLA may not be sufficient and your employer may need to sign the corporate CLA.

If you have received this in error or have any questions, please contact us at cla@fb.com. Thanks!

[facebook-github-bot]

Thank you for signing our Contributor License Agreement. We can now accept your code for this (and any) Facebook open source project. Thanks!