* Tweak command line scripts

enable fine grained control of command flags parsing
switch to attribute style of accessing config options

* Break down visualization function

decouple it from IO operations

* Move flag parsing out of `main()`

* Fix a bug where `None` is returned instead of `{}`

* Rename `save_xxx` to `output_xxx` in command line flags

could be confusing since checkpoint is stored in `save_dir`

* Support image file extensions in upper case

* Fix `norm_decay` in config files

should be float value now

* Clean up a bit

mostly max line length related

* add lstm api support; test=develop

* change name from basic_api to basic_lstm; test=develop

* fix format problem; test=develop

* add ce for dygraph mnist

* add ce for dygraph mnist

* del mnist_dygraph.py

* change mnist_dygraph to train

* fix print style

* add resnet

* fix ce bug

* fix ce decsription

* Fix ResNet18 and ResNet34 config.

* refine infer with args

* remove samples

* fix as review

* refine code

* refine args

* move get_test_images to infer.py

* add visualize log

* fix images = []

* fix args

* refine infer.py

* fix yolov3_r34.yml

* test=develop fix no random bug

* fix infer and readme

* Add ctcn model for action detection

* remove data list from codebase

developers only, not necessary for users

May be problematic for unattended runs.

* Fix type for faster_rcnn_50_2x.
* Make sure that type of cpu_num is int.

* Fix reader of slim.

* Remove unused comments.

* Fix dtype for Yolov3.
* Clean unit testing.
* Add eval() for COCO-based algo.

* Merge setup.py and requirements.txt
* Fix cascade head initializer fanout

* fix py3

To avoid the error when use_program_cache is True in Transformer infe…rence.

To avoid the data decode error.

* Unified object detection framework based on PaddlePaddle.
* Include algo: Faster, Mask, FPN, Cascade, RetinaNet, Yolo v3, SSD.

* fix grad and scipy version

* fix some bug of gan

fix op init

* fix batch generate bug in eval.py.

add AttGAN and STGAN