Reference kit update to AI Kit 2024.0.0 compatible components (#1) (#2)

* Reference kit update to AI Kit 2024.0.0 compatible components (#1)

* Remove stock environment.

* Remove env/intel directory.

* Add intel_env.yml file

* Add .gitignore

* Delete stock-pipeline.png

* Move VOC.yaml to config directory.

* Remove training.patch

* Add new training.patch

* Add output dir to .gitignore

* Move deploy.yaml to config directory.

* Add src/yolov5 to .gitignore

* Remove stock part of code from run_inference.py.

* Remove inference.png

* Delete ThroughputOpenVinoQuantizationBenchmark.png

* Update intel_env.yml.

-Add ultralytics, gitpython

* Remove data.txt

* Update README.md

* Update run_inference.py

-Changed int8inc and save_image args to store_true

* Add log info to openvino_quantization.py

* Fix logger in openvino_quantization.py

-Logger wasn't showing information at INFO level.

* Add log notifications to openvino_quantization.py

-Add 'This will take time, please wait!' message before create pipeline and execute pipeline.

* Add sections to README.md.

- Add validation tags.
- Add Expected Output.
- Add Disclaimer.
- Modify style.

* Fix README.md typos and style.

* Update .gitignore

* Update README.md

- Correct typos.
- Correct styles.
- Add rm instruction.

* Update README.md

- Correct typos.
- Correct styles.

* Update README.md

- Remove text referencing a non-existing codeblock.

* Update README.md

- Give uniform style to Notes.

* Update README.md

- Change WORKSPACE dir name.
- Change . for $WORKSPACE in git command.

* Remove traffic-detection submodule

* Update intel_env.yml

- Remove ipython from environment.

* Fix typos in README.md

* Fix dataset dir structure in README.md

* Added curl install instruction to README.md

.gitignore

@@ -0,0 +1,141 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+#   For a library or package, you might want to ignore these files since the code is
+#   intended to run in multiple environments; otherwise, check them in:
+# .python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+data/
+output/
+src/yolov5/

src/data/VOC.yaml → config/data/VOC.yaml

@@ -8,7 +8,7 @@
 # This yaml has been adapted from https://github.com/ultralytics/yolov5/blob/master/data/VOC.yaml
 
 # Train/val/test sets as 1) dir: path/to/imgs, 2) file: path/to/imgs.txt, or 3) list: [path/to/imgs1, path/to/imgs2, ..]
-path: ../../data/VOC
+path: ../data/VOC
 train: # train images (relative to 'path')  16551 images
   - images/train2012
   - images/train2007

src/training.patch → config/training.patch

@@ -1,46 +1,51 @@
 diff --git a/train.py b/train.py
-index 177e081..9b42fe9 100644
+index 4c3bec3..a82fbfb 100644
 --- a/train.py
 +++ b/train.py
-@@ -239,6 +239,16 @@ def train(hyp, opt, device, callbacks):  # hyp is path/to/hyp.yaml or hyp dictio
+@@ -24,6 +24,7 @@ import sys
+ import time
+ from copy import deepcopy
+ from datetime import datetime, timedelta
++import intel_extension_for_pytorch as ipex
+ from pathlib import Path
+
+ try:
+@@ -249,6 +250,14 @@ def train(hyp, opt, device, callbacks):  # hyp is path/to/hyp.yaml or hyp dictio
      model.class_weights = labels_to_class_weights(dataset.labels, nc).to(device) * nc  # attach class weights
      model.names = names
 
 +    # Ipex Optimizing Model
-+    if opt.intel or opt.bf16:
-+        import intel_extension_for_pytorch as ipex
-+        model.model.to(memory_format=torch.channels_last)
-+        if opt.bf16:
-+            model, optimizer = ipex.optimize(model, optimizer=optimizer, dtype=torch.bfloat16)
-+        else:
-+            model, optimizer = ipex.optimize(model, optimizer=optimizer)
-+        print("IPEX optimization enabled")
++    model.model.to(memory_format=torch.channels_last)
++    if opt.bf16:
++        model, optimizer = ipex.optimize(model, optimizer=optimizer, dtype=torch.bfloat16)
++    else:
++        model, optimizer = ipex.optimize(model, optimizer=optimizer)
++    print("IPEX optimization enabled")
 +
      # Start training
      t0 = time.time()
      nb = len(train_loader)  # number of batches
-@@ -281,7 +291,11 @@ def train(hyp, opt, device, callbacks):  # hyp is path/to/hyp.yaml or hyp dictio
+@@ -291,7 +300,10 @@ def train(hyp, opt, device, callbacks):  # hyp is path/to/hyp.yaml or hyp dictio
          for i, (imgs, targets, paths, _) in pbar:  # batch -------------------------------------------------------------
              callbacks.run('on_train_batch_start')
              ni = i + nb * epoch  # number integrated batches (since train start)
 -            imgs = imgs.to(device, non_blocking=True).float() / 255  # uint8 to float32, 0-255 to 0.0-1.0
-+
 +            if opt.bf16:
 +                imgs = imgs.to(device, non_blocking=True).bfloat16() / 255  # uint8 to float16, 0-255 to 0.0-1.0
 +            else:
 +                imgs = imgs.to(device, non_blocking=True).float() / 255  # uint8 to float32, 0-255 to 0.0-1.0
 
              # Warmup
              if ni <= nw:
-@@ -472,6 +486,11 @@ def parse_opt(known=False):
+@@ -481,6 +493,8 @@ def parse_opt(known=False):
+     parser.add_argument('--upload_dataset', nargs='?', const=True, default=False, help='Upload data, "val" option')
      parser.add_argument('--bbox_interval', type=int, default=-1, help='Set bounding-box image logging interval')
      parser.add_argument('--artifact_alias', type=str, default='latest', help='Version of dataset artifact to use')
-
-+    # Intel Ipex Optimization
-+    parser.add_argument('--intel', '-i', type=int, default=0, help='To Enable Intel Optimization set to 1, default 0')
 +    # Intel Ipex Optimization BF16
 +    parser.add_argument('--bf16', action='store_true', help='Enable only on Intel® Fourth Gen Xeon, BF16')
-+
-     return parser.parse_known_args()[0] if known else parser.parse_args()
 
+     return parser.parse_known_args()[0] if known else parser.parse_args()
 
+-- 
+2.34.1
+

env/intel_env.yml

@@ -0,0 +1,22 @@
+name: traffic_detection_intel
+channels:
+  - intel
+  - conda-forge
+dependencies:
+  - python=3.9
+  - opencv=4.8.0
+  - intel-aikit-pytorch=2024.0
+  - neural-compressor=2.3.1
+  - torchvision=0.15.2
+  - seaborn=0.13.0
+  - tqdm=4.66.1
+  - tensorboard=2.15.1
+  - pillow=10.0.1
+  - ultralytics=8.0.227
+  - gitpython=3.1.40
+  - pip=23.3.1
+  - pip:
+    - opencv-python-headless==4.8.1.78
+    - thop==0.1.1-2209072238
+    - openvino-dev[pytorch,onnx]==2023.2.0
+    - onnx==1.14.1

src/openvino/openvino_quantization.py

@@ -17,6 +17,7 @@
 import os
 from pathlib import Path
 import argparse
+import logging
 import torch
 import torch.nn.functional as nnf
 from addict import Dict
@@ -185,6 +186,18 @@ def __len__(self):
     data_path = FLAGS.data_yaml
     batch_size = FLAGS.batchsize
 
+    logger = logging.getLogger('quantization')
+    formatter = logging.Formatter(
+                "[%(asctime)s][%(levelname)s] %(message)s", "%Y-%m-%d %H:%M:%S"
+                )
+    stream_handler = logging.StreamHandler()
+    stream_handler.setFormatter(formatter)
+    stream_handler.setLevel(logging.INFO)
+    logger.addHandler(stream_handler)
+
+    logger.setLevel(logging.INFO)
+    logger.propagate = False
+
     model_config = Dict(
         {
             "model_name": "TrafficOD_Onnx_Model",
@@ -207,12 +220,14 @@ def __len__(self):
     ]
 
     # Step 1: Load the model
+    logger.info('Step 1/9: Load the model')
 
     model = load_model(model_config=model_config)
 
     original_model = copy.deepcopy(model)
 
     # Step 2: Initialize the data loader
+    logger.info('Step 2/9: Initialize the data loader')
 
     data = check_yaml(data_path)
     data = check_dataset(data)  # check
@@ -234,37 +249,47 @@ def __len__(self):
 
     data_loader = Dataset(dataloader_l)
 
-    print("This will take time, Please wait")
     # Step 3 (Optional. Required for AccuracyAwareQuantization): Initialize the metric
+    logger.info('Step 3/9: Initialize the metric')
+
     metric = Accuracy(names)
 
     # Step 4: Initialize the engine for metric calculation and statistics collection
+    logger.info('Step 4/9: Initialize the engine for metric calculation and statistics collection')
     engine = IEEngine(config=engine_config, data_loader=data_loader, metric=metric)
 
     # Step 5: Create a pipeline of compression algorithms
+    logger.info('This will take time, please wait!')
+    logger.info('Step 5/9: Create a pipeline of compression algorithms')
     pipeline = create_pipeline(algo_config=algorithms, engine=engine)
 
     original_metric_results = pipeline.evaluate(original_model)
 
     # Step 6: Execute the pipeline
+    logger.info('This will take time, please wait!')
+    logger.info('Step 6/9: Execute the pipeline')
     compressed_model = pipeline.run(model=model)
 
     # Step 7 (Optional): Compress model weights quantized precision
     #                    in order to reduce the size of final .bin file
+    logger.info('Step 7/9: Compress model weights quantized precision in" \
+                    " order to reduce the size of final .bin file')
     compress_model_weights(model=compressed_model)
 
     # Step 8: Save the compressed model and get the path to the model
+    logger.info('Step 8/9: Save the compressed model and get the path to the model')
     compressed_model_paths = save_model(
         model=compressed_model, save_path=os.path.join(os.path.curdir, out_path)
     )
     compressed_model_xml = Path(compressed_model_paths[0]["model"])
-    print(f"The quantized model is stored in {compressed_model_xml}")
+    logger.info(f"The quantized model is stored in {compressed_model_xml}")
 
     # Step 9 (Optional): Evaluate the original and compressed model. Print the results
+    logger.info('Step 9 (Optional): Evaluate the original and compressed model. Print the results')
 
     quantized_metric_results = pipeline.evaluate(compressed_model)
     if quantized_metric_results:
-        print(f"MeanAP of the quantized model: {next(iter(quantized_metric_results.values())):.5f}")
+        logger.info(f"MeanAP of the quantized model: {next(iter(quantized_metric_results.values())):.5f}")
 
     if original_metric_results:
-        print(f"MeanAP of the original model:  {next(iter(original_metric_results.values())):.5f}")
+        logger.info(f"MeanAP of the original model:  {next(iter(original_metric_results.values())):.5f}")