3
78%
master: 86%

Ran 19 Oct 2020 08:32PM UTC

Files 52

Run time 1min

Badge

Embed ▾

Committed 19 Oct 2020 08:19PM UTC coverage: 72.818%. First build

Job # TF_VERSION=2.2.1

Build Type

push

travis-ci-com

Committed by

web-flow

Commit Message

Update RetinaNet models for TensorFlow 2.2+ compatibility (#438)

* don't install tensorflow_addons (yet) and keep tf < 2.3.0

* try to test 2.3.0, hopefully can resolve this before migration.

* supporting 2.1.1+, no more py3 tags.

* no boxes_input if inclue_masks is False

* simplify shape/type dictionary instantiation

* fix mask flexibility for 3D RetinaNet

* remove deprecated "median" norm method.

* only add masks if required

* semantic heads start with the keyword

* always create semantic outputs, but it is empty if not panoptic

* watershed without passing `bins` has 1 output channel.

* add the generator output type/shape for semantic heads

* simplify `temp_boxes` and use `K.ones` instead of `tf.ones`.

* use `deepcell_toolbox.retinamask_semantic_postprocess` in notebook.

* rerun all notebooks with TF 2.3

* `pytest` 6.x does not play well with `pytest-pep8`

* new expected size due to removal of final detection layer and remove TODO comment.

* Add Python 3.6 to supported versions and upgrade TensorFlow to patched versions

* supporting TensorFlow 2.2.1+

Run Details

5090 of 6990 relevant lines covered (72.82%)

0.73 hits per line

vanvalenlab / deepcell-tf / 2142 / 3
78%
master: 86%

README BADGES
x

Markdown

Textile

RDoc

HTML

Rst

Source Files on job 2142.3 (TF_VERSION=2.2.1)

vanvalenlab / deepcell-tf / 2142 / 3 78% master: 86%

README BADGES x

Markdown

Textile

RDoc

HTML

Rst

Source Files on job 2142.3 (TF_VERSION=2.2.1)

vanvalenlab / deepcell-tf / 2142 / 3
78%
master: 86%

README BADGES
x