Python-3 support

Hi @ruotianluo , I looked at the coco-caption codebase. It seems that we need some modifications like xrange -> range etc., to port the code to Python-3.

So, I’d like to know if you have any plans to port the pycoco tools to Python-3 so that we can use this current codebase for training models in Python-3?

[salaniz]

I have implemented Python 3 support for the coco caption metrics: https://github.com/salaniz/pycocoevalcap

I decided against a PR and in favor of a fork of my own because the pycocotools that come with the original repository have diverged with the newest version present in the coco API repository here: https://github.com/cocodataset/cocoapi

Therefore, I am using the pycocotools from the API repository together with my updated caption evaluation code.

You can have a look at the changes. They are all pretty obvious, except maybe the Meteor one which took me a while to debug. If you have any problems, let me know. Also feel free to make a PR with my changes. I have not tested if my code is backward compatible with Python 2.7.

[salaniz]

My repository simply adds Python 3 support, but leaves everything else unchanged. So it is really up to how well the metrics and the tokenizer from the original coco-caption repo generalize to the Chinese language. I do not know if that’s the case.