Remove most hacks that were added to work around issues in pycocotools (

#449)

pycocotools 2.0.4 includes changes to the build system that make it possible
to downgrade NumPy after installing pycocotools, and therefore:

* --no-binary=pycocotools is no longer necessary;
* TensorFlow can be installed in the same `pip` invocation as Datumaro.

The only hack that remains (that I know of) is the usage of
`pycocotools-windows` on Windows, since the `pycocotools` PyPI project still
doesn't provide Windows wheels (or any other wheels).

.github/workflows/health_check.yml

@@ -18,8 +18,7 @@ jobs:
           python-version: ${{ matrix.python-version }}
       - name: Installing dependencies
         run: |
-          pip install tensorflow pytest pytest-cov
-          pip install -e .[default,tfds]
+          pip install -e '.[default,tf,tfds]' pytest pytest-cov
       - name: Code instrumentation
         run: |
           pytest -v --cov --cov-report xml:coverage.xml

.github/workflows/pr_checks.yml

@@ -29,8 +29,7 @@ jobs:
           python-version: ${{ matrix.python-version }}
       - name: Installing dependencies
         run: |
-          pip install tensorflow pytest
-          pip install -e .[default,tfds]
+          pip install -e '.[default,tf,tfds]' pytest
       - name: Unit testing
         run: |
           pytest -v

CHANGELOG.md

@@ -21,6 +21,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - `smooth_line` from `datumaro.util.annotation_util` - the function
   is renamed to `approximate_line` and has updated interface
   (<https://github.com/openvinotoolkit/datumaro/pull/592>)
+- The `pycocotools` dependency lower bound is raised to `2.0.4`.
+  (<https://github.com/openvinotoolkit/datumaro/pull/449>)
 
 ### Deprecated
 - TBD

requirements-core.txt

@@ -8,7 +8,7 @@ Pillow>=6.1.0
 ruamel.yaml>=0.17.0
 typing_extensions>=3.7.4.3
 
-pycocotools>=2.0.2; platform_system != "Windows" or python_version >= '3.9'
+pycocotools>=2.0.4; platform_system != "Windows" or python_version >= '3.9'
 
 pycocotools-windows; platform_system == "Windows" and python_version < '3.9'
 PyYAML>=5.3.1

requirements.txt

@@ -1,4 +1,4 @@
--r requirements-core.txt --no-binary=pycocotools # https://github.com/openvinotoolkit/datumaro/issues/253
+-r requirements-core.txt
 -r requirements-default.txt
 
 opencv-python-headless>=4.1.0.25

site/content/en/docs/user-manual/installation.md

@@ -61,20 +61,6 @@ plugin dependencies in the [plugins](/docs/user-manual/extending) section.
   This option can't be covered by extras due to Python packaging system
   limitations.
 
-- Although Datumaro excludes `pycocotools` of version 2.0.2 in
-  requirements, it works with this version perfectly fine. The
-  reason for such requirement is binary incompatibility of the `numpy`
-  dependency in the `TensorFlow` and `pycocotools` binary packages,
-  and the current workaround forces this package to be build from sources
-  on most platforms
-  (see [#253](https://github.com/openvinotoolkit/datumaro/issues/253)).
-  If you need to use 2.0.2, make sure it is linked with the same version
-  of `numpy` as `TensorFlow` by reinstalling the package:
-  ``` bash
-  pip uninstall pycocotools
-  pip install pycocotools --no-binary=pycocotools
-  ```
-
 - When installing directly from the repository, you can change the
   installation branch with `...@<branch_name>`. Also use `--force-reinstall`
   parameter in this case. It can be useful for testing of unreleased