Merge pull request #1636 from openvinotoolkit/release

Merge OpenVINO toolkit 2021.1 content into master

Author: vladimir-dudnik

.editorconfig

@@ -1,21 +1,7 @@
 root = true
 
 [*]
-end_of_line = lf
 indent_size = 4
 indent_style = space
 insert_final_newline = true
 trim_trailing_whitespace = true
-
-# 4 space indentation
-#[*.py]
-#indent_style = space
-#indent_size = 4
-
-#[{CMakeLists.txt,*.cmake}]
-#indent_style = space
-#indent_size = 4
-
-#[*.{c,cpp,h,hpp}]
-#indent_style = space
-#indent_size = 4

CONTRIBUTING.md

@@ -2,7 +2,7 @@
 
 We appreciate your intention to contribute model to the OpenVINO™ Open Model Zoo (OMZ). OMZ is licensed under the Apache\* License, Version 2.0. By contributing to the project, you agree to the license and copyright terms therein and release your contribution under these terms. Note that we accept models under permissive licenses, such as **MIT**, **Apache 2.0**, and **BSD-3-Clause**. Otherwise, it might take longer time to get your model approved.
 
-Frameworks supported by the Open Model Zoo: 
+Frameworks supported by the Open Model Zoo:
 * Caffe\*
 * Caffe2\* (via conversion to ONNX\*)
 * TensorFlow\*
@@ -45,7 +45,7 @@ File | Destination
 ---|---
 configuration file | `models/public/<model_name>/model.yml`
 documentation file | `models/public/<model_name>/<model_name>.md`
-validation configuration file|`tools/accuracy_checker/configs/<model_name>.yml`
+validation configuration file|`models/public/<model_name>/accuracy-check.yml`
 demo|`demos/<demo_name>`<br>or<br>`demos/python_demos/<demo_name>`
 
 ### Tests
@@ -66,13 +66,13 @@ Your PR may be rejected in some cases, for example:
 
 ## Configuration File
 
-The model configuration file contains information about model: what it is, how to download it, and how to convert it to the IR format. This information must be specified in the `model.yml` file that must be located in the model subfolder. 
+The model configuration file contains information about model: what it is, how to download it, and how to convert it to the IR format. This information must be specified in the `model.yml` file that must be located in the model subfolder.
 
 The detailed descriptions of file entries provided below.
 
 **`description`**
 
-Description of the model. Must match with the description from the model [documentation](#documentation). Use [this](./ci/description_updater/documentation_updater.py) script for easy update.
+Description of the model. Must match with the description from the model [documentation](#documentation). Use [this](./ci/documentation_updater/documentation_updater.py) script for easy update.
 
 **`task_type`**
 
@@ -90,7 +90,7 @@ Downloadable files. Each file is described by:
 * `source` - sets a direct link to a file *OR* describes a file access parameters
 
 > **TIP**: You can obtain a hash sum using the `sha256sum <file_name>` command on Linux\*.
- 
+
 If file is located on Google Drive\*, the `source` section must contain:
 - `$type: google_drive`
 - `id` file ID on Google Drive\*
@@ -177,21 +177,21 @@ license: https://raw.githubusercontent.com/pudae/tensorflow-densenet/master/LICE
 
 ## Model Conversion
 
-Deep Learning Inference Engine (IE) supports models in the Intermediate Representation (IR) format. A model from any supported framework can be converted to IR using the Model Optimizer tool included in the OpenVINO&trade; toolkit. Find more information about conversion in the [Model Optimizer Developer Guide](https://docs.openvinotoolkit.org/latest/_docs_MO_DG_Deep_Learning_Model_Optimizer_DevGuide.html). After a successful conversion you get a model in the IR format, with the `*.xml` file representing the net graph and the `*.bin` file containing the net parameters. 
+Deep Learning Inference Engine (IE) supports models in the Intermediate Representation (IR) format. A model from any supported framework can be converted to IR using the Model Optimizer tool included in the OpenVINO&trade; toolkit. Find more information about conversion in the [Model Optimizer Developer Guide](https://docs.openvinotoolkit.org/latest/_docs_MO_DG_Deep_Learning_Model_Optimizer_DevGuide.html). After a successful conversion you get a model in the IR format, with the `*.xml` file representing the net graph and the `*.bin` file containing the net parameters.
 
 > **NOTE 1**: Image preprocessing parameters (mean and scale) must be built into a converted model to simplify model usage.
 
 > **NOTE 2**: If a model input is a color image, color channel order should be `BGR`.
 
 ## Demo
 
-A demo shows the main idea of how to infer a model using IE. If your model solves one of the tasks supported by the Open Model Zoo, try to find an appropriate option from [demos](demos/README.md) or [samples](https://docs.openvinotoolkit.org/latest/_docs_IE_DG_Samples_Overview.html). Otherwise, you must provide your own demo (C++ or Python). 
+A demo shows the main idea of how to infer a model using IE. If your model solves one of the tasks supported by the Open Model Zoo, try to find an appropriate option from [demos](demos/README.md) or [samples](https://docs.openvinotoolkit.org/latest/_docs_IE_DG_Samples_Overview.html). Otherwise, you must provide your own demo (C++ or Python).
 
 The demo's name should end with `_demo` suffix to follow the convention of the project.
 
 Demos are required to support the following keys:
 
- -  `-i "<input>"`: Required. Input to process.
+ -  `-i "<input>"`: Required. An input to process. The input can usually be a single image, a folder of images or anything that OpenCV's `VideoCapture` can process.
  -  `-m "<path>"`: Required. Path to an .xml file with a trained model. If the demo uses several models at the same time, use other keys prefixed with `-m_`.
  -  `-d "<device>"`: Optional. Specifies a target device to infer on. CPU, GPU, FPGA, HDDL or MYRIAD is acceptable. Default must be CPU. If the demo uses several models at the same time, use keys prefixed with `d_` (just like keys `m_*` above) to specify device for each model.
  -  `-no_show`: Optional. Do not visualize inference results.
@@ -206,29 +206,18 @@ Add `README.md` file, which describes demo usage. Update [demos' README.md](demo
 
 Accuracy validation can be performed by the [Accuracy Checker](./tools/accuracy_checker) tool. This tool can use either IE to run a converted model, or an original framework to run an original model. Accuracy Checker supports lots of datasets, metrics and preprocessing options, which simplifies validation if a task is supported by the tool. You only need to create a configuration file that contains necessary parameters for accuracy validation (specify a dataset and annotation, pre- and post-processing parameters, accuracy metrics to compute and so on) of converted model. For details, refer to [Testing new models](./tools/accuracy_checker#testing-new-models).
 
-If a model uses a dataset which is not supported by the Accuracy Checker, you also must provide the license and the link to it and mention it in the PR description. 
+If a model uses a dataset which is not supported by the Accuracy Checker, you also must provide the license and the link to it and mention it in the PR description.
 
 When the configuration file is ready, you must run the Accuracy Checker to obtain metric results. If they match your results, that means conversion was successful and the Accuracy Checker fully supports your model, metric and dataset. Otherwise, recheck the [conversion](#model-conversion) parameters or the validation configuration file.
 
 ### Example
 
-This example uses one of the files from `tools/accuracy_checker/configs` — validation configuration file for [DenseNet-121](tools/accuracy_checker/configs/densenet-121-tf.yml)\* from TensorFlow\*:
+This example uses validation configuration file for [DenseNet-121](models/public/densenet-121-tf/accuracy-check.yml)\* from TensorFlow\*:
 ```
 models:
   - name: densenet-121-tf
     launchers:
       - framework: dlsdk
-        tags:
-          - FP32
-        model:   public/densenet-121-tf/FP32/densenet-121-tf.xml
-        weights: public/densenet-121-tf/FP32/densenet-121-tf.bin
-        adapter: classification
-
-      - framework: dlsdk
-        tags:
-          - FP16
-        model:   public/densenet-121-tf/FP16/densenet-121-tf.xml
-        weights: public/densenet-121-tf/FP16/densenet-121-tf.bin
         adapter: classification
 
     datasets:

README.md

@@ -1,5 +1,5 @@
 # [OpenVINO™ Toolkit](https://01.org/openvinotoolkit) - Open Model Zoo repository
-[![Stable release](https://img.shields.io/badge/version-2020.4-green.svg)](https://github.com/opencv/open_model_zoo/releases/tag/2020.4)
+[![Stable release](https://img.shields.io/badge/version-2021.1-green.svg)](https://github.com/opencv/open_model_zoo/releases/tag/2021.1)
 [![Gitter chat](https://badges.gitter.im/gitterHQ/gitter.png)](https://gitter.im/open_model_zoo/community)
 [![Apache License Version 2.0](https://img.shields.io/badge/license-Apache_2.0-green.svg)](LICENSE)
 

ci/check-release-readiness.py

@@ -0,0 +1,55 @@
+#!/usr/bin/env python3
+
+"""
+This script does various consistency checks that should pass before a release
+but that can't pass in every commit due to process imperfections.
+
+Specifically, we publish Intel models separately from their documentation and
+demo support code. Because of that, we can have situations where a model's
+config file is added, but documentation isn't, or vice versa. These need to be
+rectified before a release, so the script will warn about them.
+
+Also, we can only add a model to models.lst files once its configs are added.
+In order to not forget to do that, we can add TODO comments and the script
+will warn if any models.lst files contain such comments.
+"""
+
+import re
+import sys
+
+from pathlib import Path
+
+OMZ_ROOT = Path(__file__).resolve().parent.parent
+
+RE_TODO_COMMENT = re.compile(r'#.*\b(?:TODO|FIXME)\b')
+
+def main():
+    all_passed = True
+
+    def complain(format, *args):
+        nonlocal all_passed
+        print(format.format(*args), file=sys.stderr)
+        all_passed = False
+
+    for model_dir in OMZ_ROOT.glob('models/*/*/'):
+        # searching recursively, because this could be a composite model
+        has_config = bool(list(model_dir.glob('**/model.yml')))
+
+        has_doc = bool(list(model_dir.glob('**/*.md')))
+
+        if has_config and not has_doc:
+            complain('model {} has no documentation', model_dir.name)
+
+        if has_doc and not has_config:
+            complain('model {} has no config file', model_dir.name)
+
+    for models_lst_path in OMZ_ROOT.glob('demos/**/models.lst'):
+        with models_lst_path.open() as models_lst:
+            for line in models_lst:
+                if RE_TODO_COMMENT.search(line):
+                    complain('{} has a TODO comment', models_lst_path.relative_to(OMZ_ROOT))
+
+    sys.exit(0 if all_passed else 1)
+
+if __name__ == '__main__':
+    main()

ci/get-jobs-for-changes.py

@@ -44,7 +44,7 @@ def main():
     args = parser.parse_args()
 
     git_diff_output = subprocess.check_output(
-        ["git", "diff", "--name-only", "-z", args.base_commit + "...HEAD"])
+        ["git", "diff", "--name-only", "--no-renames", "-z", args.base_commit + "...HEAD"])
     changed_files = list(map(PurePosixPath, git_diff_output.decode()[:-1].split("\0")))
 
     models_dir = PurePosixPath("models")
@@ -55,6 +55,9 @@ def main():
         if models_dir in changed_file.parents and changed_file.name == "model.yml":
             if Path(changed_file).exists(): # it might've been deleted in the branch
                 jobs.setdefault("models", []).append(changed_file.parent.name)
+            else:
+                # make sure no models.lst files reference the deleted model
+                jobs["models_lst"] = True
 
     git_check_attr_output = subprocess.run(
         ["git", "check-attr", "--stdin", "-z", "--all"],

ci/requirements-ac-test.txt

@@ -1,42 +1,41 @@
 # use update-requirements.py to update this file
 
-atomicwrites==1.3.0       # via -r tools/accuracy_checker/requirements-test.in
+atomicwrites==1.4.0       # via -r tools/accuracy_checker/requirements-test.in
 attrs==19.3.0             # via pytest
 cycler==0.10.0            # via matplotlib
 decorator==4.4.2          # via networkx
 editdistance==0.5.3       # via -r tools/accuracy_checker/requirements.in
-imageio==2.8.0            # via scikit-image
-importlib-metadata==1.6.0  # via pluggy, pytest
+imageio==2.9.0            # via scikit-image
+importlib-metadata==1.7.0  # via pluggy, pytest
 joblib==0.14.1            # via scikit-learn
 kiwisolver==1.1.0         # via matplotlib
 matplotlib==3.0.3         # via scikit-image
-more-itertools==8.2.0     # via pytest
+more-itertools==8.4.0     # via pytest
 networkx==2.4             # via scikit-image
 nibabel==3.0.2            # via -r tools/accuracy_checker/requirements.in
 numpy==1.17.5             # via -r tools/accuracy_checker/requirements-core.in, imageio, matplotlib, nibabel, pywavelets, scikit-learn, scipy
-packaging==20.3           # via pytest
+packaging==20.4           # via pytest
 pathlib2==2.3.5           # via pytest
-pillow==7.1.1             # via -r tools/accuracy_checker/requirements.in, imageio, scikit-image
+pillow==7.2.0             # via -r tools/accuracy_checker/requirements.in, imageio, scikit-image
 pluggy==0.13.1            # via pytest
 py-cpuinfo==4.0.0         # via -r tools/accuracy_checker/requirements.in
-py==1.8.1                 # via pytest
-pydicom==1.4.2            # via -r tools/accuracy_checker/requirements.in
+py==1.9.0                 # via pytest
+pydicom==2.0.0            # via -r tools/accuracy_checker/requirements.in
 pyparsing==2.4.7          # via matplotlib, packaging
 pytest-mock==2.0.0        # via -r tools/accuracy_checker/requirements-test.in
-pytest==5.4.1             # via -r tools/accuracy_checker/requirements-test.in, pytest-mock
+pytest==5.4.3             # via -r tools/accuracy_checker/requirements-test.in, pytest-mock
 python-dateutil==2.8.1    # via matplotlib
 pywavelets==1.1.1         # via scikit-image
-pyyaml==5.3.1             # via -r tools/accuracy_checker/requirements-core.in, yamlloader
+pyyaml==5.3.1             # via -r tools/accuracy_checker/requirements-core.in
 scikit-image==0.15.0      # via -r tools/accuracy_checker/requirements.in
 scikit-learn==0.22.2.post1  # via -r tools/accuracy_checker/requirements.in
 scipy==1.4.1              # via -r tools/accuracy_checker/requirements.in, scikit-image, scikit-learn
-sentencepiece==0.1.85     # via -r tools/accuracy_checker/requirements.in
+sentencepiece==0.1.91     # via -r tools/accuracy_checker/requirements.in
 shapely==1.7.0            # via -r tools/accuracy_checker/requirements.in
-six==1.14.0               # via packaging, pathlib2, python-dateutil
-tokenizers==0.7           # via -r tools/accuracy_checker/requirements.in
-tqdm==4.45.0              # via -r tools/accuracy_checker/requirements.in
-wcwidth==0.1.9            # via pytest
-yamlloader==0.5.5         # via -r tools/accuracy_checker/requirements.in
+six==1.15.0               # via packaging, pathlib2, python-dateutil
+tokenizers==0.8.0         # via -r tools/accuracy_checker/requirements.in
+tqdm==4.47.0              # via -r tools/accuracy_checker/requirements.in
+wcwidth==0.2.5            # via pytest
 zipp==1.2.0               # via importlib-metadata
 
 # The following packages are considered to be unsafe in a requirements file:

ci/requirements-ac.txt

@@ -3,29 +3,28 @@
 cycler==0.10.0            # via matplotlib
 decorator==4.4.2          # via networkx
 editdistance==0.5.3       # via -r tools/accuracy_checker/requirements.in
-imageio==2.8.0            # via scikit-image
+imageio==2.9.0            # via scikit-image
 joblib==0.14.1            # via scikit-learn
 kiwisolver==1.1.0         # via matplotlib
 matplotlib==3.0.3         # via scikit-image
 networkx==2.4             # via scikit-image
 nibabel==3.0.2            # via -r tools/accuracy_checker/requirements.in
 numpy==1.17.5             # via -r tools/accuracy_checker/requirements-core.in, imageio, matplotlib, nibabel, pywavelets, scikit-learn, scipy
-pillow==7.1.1             # via -r tools/accuracy_checker/requirements.in, imageio, scikit-image
+pillow==7.2.0             # via -r tools/accuracy_checker/requirements.in, imageio, scikit-image
 py-cpuinfo==4.0.0         # via -r tools/accuracy_checker/requirements.in
-pydicom==1.4.2            # via -r tools/accuracy_checker/requirements.in
+pydicom==2.0.0            # via -r tools/accuracy_checker/requirements.in
 pyparsing==2.4.7          # via matplotlib
 python-dateutil==2.8.1    # via matplotlib
 pywavelets==1.1.1         # via scikit-image
-pyyaml==5.3.1             # via -r tools/accuracy_checker/requirements-core.in, yamlloader
+pyyaml==5.3.1             # via -r tools/accuracy_checker/requirements-core.in
 scikit-image==0.15.0      # via -r tools/accuracy_checker/requirements.in
 scikit-learn==0.22.2.post1  # via -r tools/accuracy_checker/requirements.in
 scipy==1.4.1              # via -r tools/accuracy_checker/requirements.in, scikit-image, scikit-learn
-sentencepiece==0.1.85     # via -r tools/accuracy_checker/requirements.in
+sentencepiece==0.1.91     # via -r tools/accuracy_checker/requirements.in
 shapely==1.7.0            # via -r tools/accuracy_checker/requirements.in
 six==1.15.0               # via python-dateutil
-tokenizers==0.7           # via -r tools/accuracy_checker/requirements.in
-tqdm==4.45.0              # via -r tools/accuracy_checker/requirements.in
-yamlloader==0.5.5         # via -r tools/accuracy_checker/requirements.in
+tokenizers==0.8.0         # via -r tools/accuracy_checker/requirements.in
+tqdm==4.47.0              # via -r tools/accuracy_checker/requirements.in
 
 # The following packages are considered to be unsafe in a requirements file:
 # setuptools