* Fix bug of FCOS when image has no positive sample

* somes changes for fcos

* Allowing validation dataset for computing validation loss

* added validation dataset

* fixing datasets errors

* Fixing linting errors

* Added vertical flipping in bbox_flip

* Fixed linting errors

* Fixed linting errors

* Added a blank line in docstrings

* Fixed linting errors.

* Add robust detection benchmark

* Update readmes

* Changed readmes for pull request

* Ensure pep8 conformity

* fixed formatting

* Fix yapf errors

* minor formatting

* fix imports order

* bug fix for recursive imports

* minor formatting

* fix the imports in __init__.py

* update contribution guide

* add isort config

* use isort to sort imports

* add isort to travis

* add major components for training double head

double head only with two losses and no attention

remove double_head detector for now, merge upchannel to double_head bbox head

change the stype using yapl

remove uncessary comment#

to pass check

to pass v2

to pass v3

line too long and style again

reuse bottlenet

* refactoring

* bug fix

* bug fix for weight initialization

* add reg roi scale factor and modify loss weights

* rescale the roi after mapping to fpn levels

* format the codebase with yapf

* add yapf to travis

* restructure the ops directory

* add some repr strings

* added cityscapes

* updated configs

* removed wip configs

* Add initial dataset instructions

* Add cityscapes readme

* Add explanation for lr scaling

* Ensure pep8 conformity

* Add CityscapesDataset to the registry

* add benchmark

* rename config, modify README.md

* fix typo

* fix typo in config

* modify INSTALL.md

Update information how to arrange cityscapes data.

* Add cityscapes class names

* Require an operation(Save or Show the results) when testing

* using parentheses

* Warning when skip image

* added parameter skip_img_without_anno for dataset

* print filename when warning

* Added Registry use for validation datasets (distributed)

* Allowing for default_args for build_dataset

* Using build_dataset instead of build_from_cfg

* Added an option to save inference results to the json file

* flake8 fixes

* yapf fixes

* Add option to use linear scaling rule

* Ensure pep8 conformity

* Add exceptions handling while validating

* fix lint

* add Exception type

* checkpoint files added

* gitignore updated with pytorch checkpoints

* Require an operation(Save or Show the results) when testing

* using parentheses

* Inference API support video

* Using mmcv.VideoReader to read video

* modify

* Fixing build-errors on Windows (add some compile args, which are default on Linux)

* Fixing build-errors on Windows

* minor formatting

* Fixing errors on windows

1.Adding 'cxx' key for extra_compile_args
2.Adding type-cast to long for first parameter of THCCeilDiv facebookresearch/maskrcnn-benchmark#409
3.Ignoring resource setting on windows

* update the order of import statements

* Implement all extension building in main setup.py

* Remove old build files

* Adapt installation guide

* Refactor to comply with flake8

* Move imports into functions & refactor module paths

* Format setup.py with yapf

* Move cython/numpy import back to head of file

* Add note to use verbose mode for pip in INSTALL.md