Cherry pick gan

* fix pickle.load in ctcn_reader for python3 (#2679)

fix pickle.load to fit both python3 and 2

* fix subprocess not stop when press ctr+c to stop main process (#2681)

* add cuda check to show errors when run on paddle compiled with cpu only (#2732)

* add cuda check to show errors when run on paddle compiled with cpu only

* change use_cuda to use_gpu in the error infomation to be consistent with code

* Fix some docs. (#2698)
* Unify COCO and VOC
* Change PASCAL to Pascal
* Unify dataset/coco

* [cherry pick] fix configs in paddle detection

* [cherry pick] add depth for ResNetC5

* improve quality of images

* update url for install doc

* add readme for ctcn metrics calculation

* fix code format in metrics/detections/README.md

* fix code format and add ``` in code lines

* Upload mmpms

* Adapt to py2

* change seq2seq to padding impl; test=develop

* add bleu result; test=develop

* fix formate; test=develop

* fix formate; test=develop

* Reduce buf size in image_classification/reader
* Fix format

* refine README & use a better visualization for doc

* fix typo in GETTING STARTED

* Update README.md and GETTING_STARTED.md for PaddleDetection.

* fix description of video in models/readme to be same as PaddleVideo/readme
* add c2d and i3d backbone in nonloal/readme
* fix table format of nonlocal/readme

* fix model zoo doc

* update yolov3/ssd bbox AP

* refine MODEL ZOO

* fix AP

* add size 416 320

* fix link

* fix format

* fix align

* add model zoo document file

* polish and fix some errors in README.md

* Update README.md

* Update README.md

* refine readme and fix train bug

* update darknet53 pretrain link

* update weights path