AI 深度學習與視覺運算 –以OpenVINO 和 Phython 實例介紹

# AI 深度學習與視覺運算 –以OpenVINO 和 Phython 實例介紹 ## 安裝 OpenVINO Development Tools (PypI package) [Installing OpenVINO™ Development Tools](https://docs.openvino.ai/latest/openvino_docs_install_guides_install_dev_tools.html#install-dev-tools) Step 1. 設定 Python 虛擬環境 ``` python -m venv openvino_env ``` Step 2. 啟動虛擬環境 ``` Scripts\activate ``` Step 3. Set Up and Update PIP to the Highest Version ``` python -m pip install --upgrade pip ``` Step 4. Install the Package ``` pip install openvino-dev[tensorflow2,onnx,pytorch] ``` Step 5. Test the Installation ``` mo -h ``` --- ## 安裝 Jupyter Notebook ``` pip install jupyter ``` --- ## 執行 Jupyter Notebook 下載[00 OpenVINO Quick Get Started.ipynb](https://drive.google.com/drive/folders/1lLvl45Y2E8z8yiR16M8efsgr3giC9vzb?usp=sharing)到C:\Users\\<font color="#f00">xxx</font>\openvino_env 路徑下在虛擬環境下執行 ``` jupyter notebook ``` [car](https://drive.google.com/file/d/1wFZbFWmzh6iDgr1G8H-JeD17Quwn3LTs/view?usp=share_link) --- ## 如何進入OpenVINO 開發虛擬環境搜尋"命令列" or "command prompt" 並執行啟動虛擬環境 ``` Scripts\activate ``` --- ## Pre-trained Models [Intel’s Pre-Trained Models ](https://docs.openvino.ai/latest/omz_models_group_intel.html ) [Public Pre-Trained Models ](https://docs.openvino.ai/latest/omz_models_group_intel.html ) --- ## Demo Applications [Demo Applications ](https://docs.openvino.ai/latest/omz_demos.html ) --- ## 下載Open Model Zoo 由以下網址[Open Model Zoo](https://github.com/openvinotoolkit/open_model_zoo/archive/refs/heads/releases/2022/2.zip ), 直接下載 2022.2 release 壓縮檔把下載的open_model_zoo-releases-2022-2解壓縮到 , 比如解壓縮到c:\users\\<font color="#f00">xxx</font>\open_model_zoo. 執行 ``` Scripts\activate ``` 執行 ``` pip install -r \open_model_zoo\demos\requirements.txt ``` 測試 Python Demo是否可以執行切換目錄到 cd \open_model_zoo\demos\object_detection_demo\python 執行 ``` set PYTHONPATH=%PYTHONPATH%;c:\Users\colli\open_model_zoo\demos\common\python\ ``` 執行 ``` python object_detection_demo.py -h ``` --- ## 執行影像分類範例 (classification demo) 準備工作設定python API 路徑 ``` set PYTHONPATH=%PYTHONPATH%;c:\Users\colli\open_model_zoo\demos\common\python\ ``` 切換目錄到 ``` cd C:\Users\colli\open_model_zoo\demos\classification_demo\python ``` 下載與轉換深度學習模型到 IR models ``` omz_downloader --name googlenet-v1 omz_converter --name googlenet-v1 ``` 執行範例程式 ``` python classification_demo.py -m public\googlenet-v1\FP16\googlenet-v1.xml -i dog.bmp -d GPU ``` 加入label ``` python classification_demo.py -m public\googlenet-v1\FP16\googlenet-v1.xml -i dog.bmp -d GPU --labels c:\Users\colli\open_model_zoo\data\dataset_classes/imagenet_2012.txt ``` [蔬菜水果分類範例影片 ](https://storage.openvinotoolkit.org/test_data/videos/fruit-and-vegetable-detection.mp4 ) [Classification python demo ](https://docs.openvino.ai/latest/omz_demos_classification_demo_python.html ) [dog](https://drive.google.com/file/d/1U9mQHvxhYRZ6kZBpxLxjmwtWLNX3CRF2/view?usp=share_link) --- ## Python classification_demo.py -h ``` Python classification_demo.py -h ``` --- ## 執行Object Detection Demo (SSD) 切換目錄到 ``` cd C:\Users\colli\open_model_zoo\demos\object_detection_demo\python ``` 下載與轉換深度學習模型到 IR models SSD300 ``` omz_downloader --name ssd300 omz_converter --name ssd300 ``` 執行範例程式 (using USB webcam as input) ``` python object_detection_demo.py -m public\ssd300\FP16\ssd300.xml -i 0 -at ssd --labels C:\Users\colli\open_model_zoo\data\dataset_classes\voc_20cl_bkgr.txt ``` ``` python object_detection_demo.py -m public\ssd300\FP16\ssd300.xml -i car-detection.mp4 -at ssd --labels C:\Users\colli\open_model_zoo\data\dataset_classes\voc_20cl_bkgr.txt ``` [汽車偵測範例影片 ](https://storage.openvinotoolkit.org/test_data/videos/car-detection.mp4 ) [Object Detection C++ Demo ](https://docs.openvino.ai/latest/omz_demos_object_detection_demo_python.html ) [ssd300 ](https://docs.openvino.ai/latest/omz_models_model_ssd300.html#doxid-omz-models-model-ssd300) --- ## How to Run Text Spotting Python* Demo 切換目錄到 ``` cd C:\Users\colli\open_model_zoo\demos\text_spotting_demo\python ``` 下載與轉換深度學習模型到 IR models ``` omz_downloader --name text-spotting-0005-detector omz_downloader --name text-spotting-0005-recognizer-encoder omz_downloader --name text-spotting-0005-recognizer-decoder ``` 執行範例程式 ``` python text_spotting_demo.py -m_m intel\text-spotting-0005\text-spotting-0005-detector\FP16\text-spotting-0005-detector.xml -m_te intel\text-spotting-0005\text-spotting-0005-recognizer-encoder\FP16\text-spotting-0005-recognizer-encoder.xml -m_td intel\text-spotting-0005\text-spotting-0005-recognizer-decoder\FP16\text-spotting-0005-recognizer-decoder.xml -i Einstein.png ``` [Text Spotting Python Demo ](https://docs.openvino.ai/latest/omz_demos_text_spotting_demo_python.html ) [text-spotting-0005 ](https://docs.openvino.ai/latest/omz_models_model_text_spotting_0005.html) [Einstein](https://drive.google.com/file/d/1N8jUoqNpcQ8yVUtkYHhBSff-d_hB0OXz/view?usp=share_link) --- ## How to Run Human Pose Estimation Python* Demo 切換目錄到 ``` cd C:\Users\colli\open_model_zoo\demos\human_pose_estimation_demo\python ``` 下載Intel深度學習模型 ``` omz_downloader --name human-pose-estimation-0005 ``` 執行範例程式 (using USB webcam as input) ``` python human_pose_estimation_demo.py -d CPU -i 0 -m intel\human-pose-estimation-0005\FP16\human-pose-estimation-0005.xml -at ae ``` 執行範例程式 using Video file ``` python human_pose_estimation_demo.py -d CPU -i face-demographics-walking.mp4 -m intel\human-pose-estimation-0005\FP16\human-pose-estimation-0005.xml -at ae ``` [Human Pose Estimation Python* Demo ](https://docs.openvino.ai/latest/omz_demos_human_pose_estimation_demo_python.html ) [human-pose-estimation-0005 ](https://docs.openvino.ai/latest/omz_models_model_human_pose_estimation_0005.html) [face-demographics-walking 影片檔](https://storage.openvinotoolkit.org/data/test_data/videos/face-demographics-walking.mp4) --- ## 安裝OpenVINO Notebooks 下載 OpenVINO Notebooks 原始碼，https://github.com/openvinotoolkit/openvino_notebooks/archive/refs/heads/main.zip Install C++ Redistributable (For Python 3.8)- Download [Microsoft Visual C++ Redistributable](https://aka.ms/vs/16/release/vc_redist.x64.exe). 把下載的openvino_notebooks-main.zip解壓縮到 , 比如解壓縮到c:\users\colli\openvino_env\ 到目錄c:\users\colli\openvino_env下執行 ``` Scripts\activate ``` 執行 ``` python -m pip install --upgrade pip wheel setuptools ``` 執行 ``` pip install -r requirements.txt ``` --- ## OpenVINO Notebooks [OpenVINO Notebooks](https://github.com/openvinotoolkit/openvino_notebooks ) --- ## Smart Classroom C++ Demo Anaconda prompt 下執行 ``` conda activate ov_2022_dev ``` ``` cd C:\Program Files (x86)\Intel\openvino_2022 setupvars.bat ``` ``` cd C:\Users\colli\Documents\Intel\OpenVINO\omz_demos_build\intel64\Release ``` ``` smart_classroom_demo -i 1 -m_act intel\person-detection-action-recognition-0006\FP16\person-detection-action-recognition-0006.xml -student_ac "sitting, writing, raising hand, standing, turned around, lie on the desk" -m_fd intel\face-detection-adas-0001\FP16\face-detection-adas-0001.xml -m_reid intel\face-reidentification-retail-0095\FP16\face-reidentification-retail-0095.xml -m_lm intel\landmarks-regression-retail-0009\FP16\landmarks-regression-retail-0009.xml -t_reid 0.8 -fg C:\Users\colli\open_model_zoo\faces_gallery.json ```