私的AI研究会 > OpenVINO
2020年12月15日現在、OpenVINO™ ツールキットの最新バージョンは 2021.2
公式サイト → Install OpenVINO™ toolkit for Raspbian* OS
● VNCビューアなどで、Raspberry Pi 上のブラウザでオフィシャルサイトのダウンロードページから下記のファイルをダウンロードする。
● URL:https://storage.openvinotoolkit.org/repositories/openvino/packages/2021.2/
● ファイル名:l_openvino_toolkit_runtime_raspbian_p_2021.2.185.tgz
※ファイルは「~/Download」ディレクトリに配置される。pi@raspberrypi:~ $ cd Downloads pi@raspberrypi:~/Downloads $ ls l_openvino_toolkit_runtime_raspbian_p_2021.2.185.tgz
pi@raspberrypi:~/Downloads $ sudo mkdir -p /opt/intel/openvino pi@raspberrypi:~/Downloads $ ls /opt/intel openvino
pi@raspberrypi:~/Downloads $ sudo tar -xf l_openvino_toolkit_runtime_raspbian_p_2021.2.185.tgz --strip 1 -C /opt/intel/openvino pi@raspberrypi:~/Downloads $ ls /opt/intel/openvino bin deployment_tools documentation inference_engine install_dependencies licensing opencv python
pi@raspberrypi:~/Downloads $ sudo apt update pi@raspberrypi:~/Downloads $ sudo apt install cmake パッケージリストを読み込んでいます... 完了 依存関係ツリーを作成しています 状態情報を読み取っています... 完了 以下の追加パッケージがインストールされます: cmake-data libjsoncpp1 librhash0 : この操作後に追加で 21.6 MB のディスク容量が消費されます。 続行しますか? [Y/n] y 取得:1 http://ftp.jaist.ac.jp/raspbian buster/main armhf cmake-data all 3.13.4-1 [1,476 kB] 取得:2 http://ftp.jaist.ac.jp/raspbian buster/main armhf libjsoncpp1 armhf 1.7.4-3 [66.2 kB] :
pi@raspberrypi:~/Downloads $ source /opt/intel/openvino/bin/setupvars.sh [setupvars.sh] OpenVINO environment initialized
pi@raspberrypi:~/Downloads $ echo "source /opt/intel/openvino/bin/setupvars.sh" >> ~/.bashrc
pi@raspberrypi:~/Downloads $ sudo usermod -a -G users "$(whoami)" pi@raspberrypi:~/Downloads $
pi@raspberrypi:~/Downloads $ sh /opt/intel/openvino/install_dependencies/install_NCS_udev_rules.sh Updating udev rules... Udev rules have been successfully installed.
pi@raspberrypi:~ $ mkdir build pi@raspberrypi:~ $ cd build
pi@raspberrypi:~/build $ cmake -DCMAKE_BUILD_TYPE=Release -DCMAKE_CXX_FLAGS="-march=armv7-a" /opt/intel/openvino/deployment_tools/inference_engine/samples/cpp -- The C compiler identification is GNU 8.3.0 -- The CXX compiler identification is GNU 8.3.0 -- Check for working C compiler: /usr/bin/cc -- Check for working C compiler: /usr/bin/cc -- works -- Detecting C compiler ABI info -- Detecting C compiler ABI info - done -- Detecting C compile features -- Detecting C compile features - done -- Check for working CXX compiler: /usr/bin/c++ -- Check for working CXX compiler: /usr/bin/c++ -- works -- Detecting CXX compiler ABI info -- Detecting CXX compiler ABI info - done -- Detecting CXX compile features -- Detecting CXX compile features - done -- Looking for C++ include unistd.h -- Looking for C++ include unistd.h - found -- Looking for C++ include stdint.h -- Looking for C++ include stdint.h - found -- Looking for C++ include sys/types.h -- Looking for C++ include sys/types.h - found -- Looking for C++ include fnmatch.h -- Looking for C++ include fnmatch.h - found -- Looking for strtoll -- Looking for strtoll - found -- Found InferenceEngine: /opt/intel/openvino/deployment_tools/inference_engine/lib/armv7l/libinference_engine.so (Required is at least version "2.1") -- Configuring done -- Generating done -- Build files have been written to: /home/pi/build
pi@raspberrypi:~/build $ make -j2 object_detection_sample_ssd Scanning dependencies of target format_reader Scanning dependencies of target gflags_nothreads_static [ 9%] Building CXX object thirdparty/gflags/CMakeFiles/gflags_nothreads_static.dir/src/gflags.cc.o [ 18%] Building CXX object common/format_reader/CMakeFiles/format_reader.dir/MnistUbyte.cpp.o [ 27%] Building CXX object common/format_reader/CMakeFiles/format_reader.dir/bmp.cpp.o [ 36%] Building CXX object common/format_reader/CMakeFiles/format_reader.dir/format_reader.cpp.o [ 45%] Building CXX object thirdparty/gflags/CMakeFiles/gflags_nothreads_static.dir/src/gflags_reporting.cc.o [ 54%] Building CXX object common/format_reader/CMakeFiles/format_reader.dir/opencv_wraper.cpp.o [ 63%] Building CXX object thirdparty/gflags/CMakeFiles/gflags_nothreads_static.dir/src/gflags_completions.cc.o [ 72%] Linking CXX static library ../../armv7l/Release/lib/libgflags_nothreads.a [ 72%] Built target gflags_nothreads_static [ 81%] Linking CXX shared library ../../armv7l/Release/lib/libformat_reader.so [ 81%] Built target format_reader Scanning dependencies of target object_detection_sample_ssd [ 90%] Building CXX object object_detection_sample_ssd/CMakeFiles/object_detection_sample_ssd.dir/main.cpp.o [100%] Linking CXX executable ../armv7l/Release/object_detection_sample_ssd [100%] Built target object_detection_sample_ssd
pi@raspberrypi:~/build $ wget --no-check-certificate https://download.01.org/opencv/2021/openvinotoolkit/2021.2/open_model_zoo/models_bin/3/face-detection-adas-0001/FP16/face-detection-adas-0001.bin --2020-12-18 11:24:17-- https://download.01.org/opencv/2021/openvinotoolkit/2021.2/open_model_zoo/models_bin/3/face-detection-adas-0001/FP16/face-detection-adas-0001.bin download.01.org (download.01.org) をDNSに問いあわせています... 23.41.94.105, 2600:140b:8800:28f::4b21, 2600:140b:8800:283::4b21 download.01.org (download.01.org)|23.41.94.105|:443 に接続しています... 接続しました。 HTTP による接続要求を送信しました、応答を待っています... 200 OK 長さ: 2106088 (2.0M) [application/octet-stream] `face-detection-adas-0001.bin' に保存中 face-detection-adas-0001.bin 100%[=======================================================>] 2.01M 2.06MB/s 時間 1.0s 2020-12-18 11:24:20 (2.06 MB/s) - `face-detection-adas-0001.bin' へ保存完了 [2106088/2106088]
pi@raspberrypi:~/build $ wget --no-check-certificate https://download.01.org/opencv/2021/openvinotoolkit/2021.2/open_model_zoo/models_bin/3/face-detection-adas-0001/FP16/face-detection-adas-0001.xml --2020-12-18 11:24:42-- https://download.01.org/opencv/2021/openvinotoolkit/2021.2/open_model_zoo/models_bin/3/face-detection-adas-0001/FP16/face-detection-adas-0001.xml download.01.org (download.01.org) をDNSに問いあわせています... 23.41.94.105, 2600:140b:8800:28f::4b21, 2600:140b:8800:283::4b21 download.01.org (download.01.org)|23.41.94.105|:443 に接続しています... 接続しました。 HTTP による接続要求を送信しました、応答を待っています... 200 OK 長さ: 225964 (221K) [text/xml] `face-detection-adas-0001.xml' に保存中 face-detection-adas-0001.xml 100%[=======================================================>] 220.67K 688KB/s 時間 0.3s 2020-12-18 11:24:44 (688 KB/s) - `face-detection-adas-0001.xml' へ保存完了 [225964/225964]
pi@raspberrypi:~/build $ ls CMakeCache.txt classification_sample_async hello_classification object_detection_sample_ssd CMakeFiles cmake_install.cmake hello_nv12_input_classification speech_sample Makefile common hello_query_device style_transfer_sample armv7l face-detection-adas-0001.bin hello_reshape_ssd thirdparty benchmark_app face-detection-adas-0001.xml ngraph_function_creation_sample
pi@raspberrypi:~/build $ ./armv7l/Release/object_detection_sample_ssd -m face-detection-adas-0001.xml -d MYRIAD -i input.jpg [ INFO ] InferenceEngine: API version ............ 2.1 Build .................. 2021.2.0-1877-176bdf51370-releases/2021/2 Description ....... API Parsing input parameters [ INFO ] Files were added: 1 [ INFO ] input.jpg [ INFO ] Loading Inference Engine [ INFO ] Device info: MYRIAD myriadPlugin version ......... 2.1 Build ........... 2021.2.0-1877-176bdf51370-releases/2021/2 [ INFO ] Loading network files: face-detection-adas-0001.xml [ INFO ] Preparing input blobs [ INFO ] Batch size is 1 [ INFO ] Preparing output blobs [ INFO ] Loading model to the device [ INFO ] Create infer request [ WARNING ] Image is resized from (1440, 1048) to (672, 384) [ INFO ] Batch size is 1 [ INFO ] Start inference [ INFO ] Processing output blobs [0,1] element, prob = 1 (615,122)-(827,377) batch id : 0 WILL BE PRINTED! [1,1] element, prob = 1 (875,378)-(1068,628) batch id : 0 WILL BE PRINTED! [2,1] element, prob = 0.0756836 (1227,2)-(1338,139) batch id : 0 : : [91,1] element, prob = 0.0356445 (1158,123)-(1199,186) batch id : 0 [92,1] element, prob = 0.0356445 (273,165)-(317,239) batch id : 0 [93,1] element, prob = 0.0356445 (294,203)-(344,288) batch id : 0 [ INFO ] Image out_0.bmp created! E: [global] [ 337810] [object_detectio] XLink_sem_wait:94 XLink_sem_inc(sem) method call failed with an error: -1 E: [global] [ 337810] [object_detectio] XLinkResetRemote:257 can't wait dispatcherClosedSem [ INFO ] Execution successful [ INFO ] This sample is an API example, for any performance measurements please use the dedicated benchmark_app tool
インテル®が公開しているトレーニング済みモデル(※2020/12/18現在)
以下のリンク先の下の /FP16 にあるファイル「XXXXXX.bin」「XXXXXX.xml」を前節4. 以降の手順で利用することができる。
https://download.01.org/opencv/2021/openvinotoolkit/2021.2/open_model_zoo/models_bin/3/
FILE NAME FILE SIZE DATE Parent directory/ - - action-recognition-0001 - 2020-Dec-10 16:54 age-gender-recognition-retail-0013 - 2020-Dec-10 16:54 asl-recognition-0004 - 2020-Dec-10 16:54 bert-large-uncased-whole-word-masking-squad-0001 - 2020-Dec-10 16:54 bert-large-uncased-whole-word-masking-squad-emb..> - 2020-Dec-10 16:54 bert-large-uncased-whole-word-masking-squad-int..> - 2020-Dec-10 16:54 bert-small-uncased-whole-word-masking-squad-0001 - 2020-Dec-10 16:54 bert-small-uncased-whole-word-masking-squad-0002 - 2020-Dec-10 16:54 bert-small-uncased-whole-word-masking-squad-emb..> - 2020-Dec-10 16:54 bert-small-uncased-whole-word-masking-squad-int..> - 2020-Dec-10 16:54 driver-action-recognition-adas-0002 - 2020-Dec-10 16:54 emotions-recognition-retail-0003 - 2020-Dec-10 16:54 face-detection-0200 - 2020-Dec-10 16:54 face-detection-0202 - 2020-Dec-10 16:54 face-detection-0204 - 2020-Dec-10 16:54 face-detection-0205 - 2020-Dec-10 16:54 face-detection-0206 - 2020-Dec-10 16:54 face-detection-adas-0001 - 2020-Dec-10 16:54 face-detection-retail-0004 - 2020-Dec-10 16:54 face-detection-retail-0005 - 2020-Dec-10 16:54 facial-landmarks-35-adas-0002 - 2020-Dec-10 16:54 faster-rcnn-resnet101-coco-sparse-60-0001 - 2020-Dec-10 16:54 formula-recognition-medium-scan-0001 - 2020-Dec-10 16:54 formula-recognition-polynomials-handwritten-0001 - 2020-Dec-10 16:54 gaze-estimation-adas-0002 - 2020-Dec-10 16:54 handwritten-japanese-recognition-0001 - 2020-Dec-10 16:54 handwritten-score-recognition-0003 - 2020-Dec-10 16:54 handwritten-simplified-chinese-recognition-0001 - 2020-Dec-10 16:54 head-pose-estimation-adas-0001 - 2020-Dec-10 16:54 horizontal-text-detection-0001 - 2020-Dec-10 16:54 human-pose-estimation-0001 - 2020-Dec-10 16:54 human-pose-estimation-0002 - 2020-Dec-10 16:54 human-pose-estimation-0003 - 2020-Dec-10 16:54 human-pose-estimation-0004 - 2020-Dec-10 16:54 icnet-camvid-ava-0001 - 2020-Dec-10 16:54 icnet-camvid-ava-sparse-30-0001 - 2020-Dec-10 16:54 icnet-camvid-ava-sparse-60-0001 - 2020-Dec-10 16:54 image-retrieval-0001 - 2020-Dec-10 16:54 instance-segmentation-security-0010 - 2020-Dec-10 16:54 instance-segmentation-security-0050 - 2020-Dec-10 16:54 instance-segmentation-security-0083 - 2020-Dec-10 16:54 instance-segmentation-security-1025 - 2020-Dec-10 16:54 landmarks-regression-retail-0009 - 2020-Dec-10 16:54 license-plate-recognition-barrier-0001 - 2020-Dec-10 16:54 machine-translation-nar-en-ru-0001 - 2020-Dec-10 16:54 machine-translation-nar-ru-en-0001 - 2020-Dec-10 16:54 pedestrian-and-vehicle-detector-adas-0001 - 2020-Dec-10 16:54 pedestrian-detection-adas-0002 - 2020-Dec-10 16:54 person-attributes-recognition-crossroad-0230 - 2020-Dec-10 16:54 person-attributes-recognition-crossroad-0234 - 2020-Dec-10 16:54 person-attributes-recognition-crossroad-0238 - 2020-Dec-10 16:54 person-detection-0106 - 2020-Dec-10 16:54 person-detection-0200 - 2020-Dec-10 16:54 person-detection-0201 - 2020-Dec-10 16:54 person-detection-0202 - 2020-Dec-10 16:54 person-detection-0203 - 2020-Dec-10 16:54 person-detection-action-recognition-0005 - 2020-Dec-10 16:54 person-detection-action-recognition-0006 - 2020-Dec-10 16:54 person-detection-action-recognition-teacher-0002 - 2020-Dec-10 16:54 person-detection-asl-0001 - 2020-Dec-10 16:54 person-detection-raisinghand-recognition-0001 - 2020-Dec-10 16:54 person-detection-retail-0002 - 2020-Dec-10 16:54 person-detection-retail-0013 - 2020-Dec-10 16:54 person-reidentification-retail-0277 - 2020-Dec-10 16:54 person-reidentification-retail-0286 - 2020-Dec-10 16:54 person-reidentification-retail-0287 - 2020-Dec-10 16:54 person-reidentification-retail-0288 - 2020-Dec-10 16:54 person-vehicle-bike-detection-2000 - 2020-Dec-10 16:54 person-vehicle-bike-detection-2001 - 2020-Dec-10 16:54 person-vehicle-bike-detection-2002 - 2020-Dec-10 16:54 person-vehicle-bike-detection-crossroad-0078 - 2020-Dec-10 16:54 person-vehicle-bike-detection-crossroad-1016 - 2020-Dec-10 16:54 person-vehicle-bike-detection-crossroad-yolov3-..> - 2020-Dec-10 16:54 product-detection-0001 - 2020-Dec-10 16:54 resnet18-xnor-binary-onnx-0001 - 2020-Dec-10 16:54 resnet50-binary-0001 - 2020-Dec-10 16:54 road-segmentation-adas-0001 - 2020-Dec-10 16:54 semantic-segmentation-adas-0001 - 2020-Dec-10 16:54 single-image-super-resolution-1032 - 2020-Dec-10 16:54 single-image-super-resolution-1033 - 2020-Dec-10 16:54 text-detection-0003 - 2020-Dec-10 16:54 text-detection-0004 - 2020-Dec-10 16:54 text-image-super-resolution-0001 - 2020-Dec-10 16:54 text-recognition-0012 - 2020-Dec-10 16:54 text-spotting-0003 - 2020-Dec-10 16:54 unet-camvid-onnx-0001 - 2020-Dec-10 16:54 vehicle-attributes-recognition-barrier-0039 - 2020-Dec-10 16:54 vehicle-attributes-recognition-barrier-0042 - 2020-Dec-10 16:54 vehicle-detection-0200 - 2020-Dec-10 16:54 vehicle-detection-0201 - 2020-Dec-10 16:54 vehicle-detection-0202 - 2020-Dec-10 16:54 vehicle-detection-adas-0002 - 2020-Dec-10 16:54 vehicle-license-plate-detection-barrier-0106 - 2020-Dec-10 16:54 weld-porosity-detection-0001 - 2020-Dec-10 16:54 yolo-v2-ava-0001 - 2020-Dec-10 16:54 yolo-v2-ava-sparse-35-0001 - 2020-Dec-10 16:54 yolo-v2-ava-sparse-70-0001 - 2020-Dec-10 16:54 yolo-v2-tiny-ava-0001 - 2020-Dec-10 16:54 yolo-v2-tiny-ava-sparse-30-0001 - 2020-Dec-10 16:54 yolo-v2-tiny-ava-sparse-60-0001 - 2020-Dec-10 16:54 yolo-v2-tiny-vehicle-detection-0001 - 2020-Dec-10 16:54 LICENSE 11.1 KiB 2020-Dec-10 16:54