GitLab

resolve dist_train issues by fixing mmcv==0.2.16

This reverts commit eda7242b.

This reverts commit 10830bc7.

fix mmcv version to resolve issue #4

* add reppoints without gn

* fix

* add logger utils

* replace get_root_logger() and logger.info() with print_log()

* fix a typo

* minor fix for the format of StreamHandler

* bump version to 1.0.0

* fix typos

* update the version in docs

* use git repo instead of pip wheels

* install pycocotools before building mmdet

* fix travis installation

* options for mask during testtime

* handle comments

* Enhance AssignResult and SamplingResult

Add runtime dependency on ubelt (pending approval)

Fix issue in SamplingResult.__init__

Add rng as attribute of RandomSampler

* fix linters

* remove ubelt

* Fix linters

* fix linters again

* reorganize requirements, make albumentations optional

* fix flake8 error

* force older version of Pillow until torchvision is fixed

* make imagecorruptions optional and update INSTALL.dm

* update INSTALL.md

* Add note about pillow version

* Add build requirements to install instructions

* update code

* reformat

* fix load pretrain bug

* reformat

* fix key warnings

* use build_conv_layer & ConvModule

* rm unrelated file

* fix resnext group bug

* change cfg to pass test

* use build_conv_layer to claim dcn in resnets

* change _version = 2

* MDCN to DCNv2

* change dependency of mmcv from 0.2.15 to 0.2.16

* resolve comments

* rm comments

* get_root_logger()

* Code for Paper "Bridging the Gap Between Anchor-based and Anchor-free Detection via Adaptive Training Sample Selection"

* fix format

* changed the code according to the Changes requested

* update benchmark

* minor refactoring

* minor fix

* update model_zoo and support models
Co-authored-by: Cao Yuhang <yhcao6@gmail.com>

* First Draft

On branch non-color-images
Changes to be committed:
modified:   mmdet/datasets/pipelines/loading.py

* Add option to load non color images

Add 'color_type' parameter to LoadImageFromFile class
Change __repr__ method accordingly
Since non-color images maybe two dimensional expand image
dimensions if necessary in DefaultFormatBundle and
ImageToTensor classes

Changes to be committed:
    modified:   mmdet/datasets/pipelines/formating.py
    modified:   mmdet/datasets/pipelines/loading.py

* Fix RandomCrop to work with grayscale images

Changes to be committed:
    modified:   mmdet/datasets/pipelines/transforms.py

* Modify retrieving w, h of padded image in anchor heads

This addreses problems with single channel images for which the
shape is  tuple with 2 values

Changes to be committed:
    modified:   mmdet/models/anchor_heads/anchor_head.py
    modified:   mmdet/models/anchor_heads/guided_anchor_head.py
    modified:   mmdet/models/anchor_heads/reppoints_head.py