[E-103104] - Port gaze_estimation_demo gapi demo to OV API2.0 (#3891)

* Port gaze_estimation_demo gapi demo to OV API2.0 * Port gaze_estimation_demo gapi demo to OV API2.0 - video input * Port gaze_estimation_demo gapi demo to OV API2.0 - review comments
openvinotoolkit · Jan 5, 2024 · 8c9bc0d · 8c9bc0d
1 parent e8fb4cd
commit 8c9bc0d
Show file tree

Hide file tree

Showing 4 changed files with 21 additions and 11 deletions.
diff --git a/.github/workflows/cpp_gapi-demos.yml b/.github/workflows/cpp_gapi-demos.yml
@@ -35,10 +35,11 @@ jobs:
       - if: ${{ !steps.cache.outputs.cache-hit }}
         name: Compile OpenCV
         run: |
+          sudo apt install pkg-config ffmpeg libavcodec-dev libavformat-dev libavutil-dev libswscale-dev libavresample-dev
           rm -rf cache/opencv/.git/  # Minimize cache
           mkdir cache/opencv/build
           cd cache/opencv/build
-          cmake -DCMAKE_BUILD_TYPE=Release -DWITH_INF_ENGINE=y -DOpenVINO_DIR=$GITHUB_WORKSPACE/ov/runtime/cmake/ -DCMAKE_CXX_COMPILER_LAUNCHER=ccache -DCMAKE_CXX_LINKER_LAUNCHER=ccache -DCMAKE_C_COMPILER_LAUNCHER=ccache -DCMAKE_C_LINKER_LAUNCHER=ccache -DBUILD_TESTS=n -DBUILD_PERF_TESTS=n -DBUILD_EXAMPLES=n -DBUILD_opencv_apps=n -DWITH_OPENCL=n -DWITH_OPENCLAMDBLAS=n -DWITH_OPENCLAMDFFT=n -DWITH_VA=n -DWITH_VA_INTEL=n -DWITH_V4L=n -DWITH_GSTREAMER=n -DWITH_PROTOBUF=n -DBUILD_PROTOBUF=n -DBUILD_JAVA=n -DBUILD_opencv_java_bindings_generator=n -DBUILD_opencv_python2=n -DBUILD_opencv_python3=n -DWITH_IMGCODEC_HDR=n -DWITH_IMGCODEC_SUNRASTER=n -DWITH_IMGCODEC_PXM=n -DWITH_IMGCODEC_PFM=n -DWITH_PNG=n -DWITH_TIFF=n -DWITH_WEBP=n -DWITH_OPENJPEG=n -DWITH_JASPER=n -DWITH_OPENEXR=n -DBUILD_opencv_dnn=n -DBUILD_opencv_features2d=n -DBUILD_opencv_flann=n -DWITH_TBB=n -DBUILD_INFO_SKIP_EXTRA_MODULES=n -DBUILD_JASPER=n -DBUILD_PNG=n -DBUILD_OPENEXR=n -DBUILD_WEBP=n -DBUILD_ZLIB=n -DWITH_CUDA=n -DWITH_EIGEN=n -DWITH_GPHOTO2=n -DOPENCV_GAPI_GSTREAMER=n -DWITH_LAPACK=n -DWITH_MATLAB=n -DWITH_MFX=n -DWITH_QUIRC=n -DWITH_VTK=n -DINSTALL_PDB=n -DINSTALL_TESTS=n -DINSTALL_C_EXAMPLES=n -DINSTALL_PYTHON_EXAMPLES=n -DOPENCV_GENERATE_SETUPVARS=n -DWITH_1394=n -DWITH_FFMPEG=n -DWITH_GTK_2_X=y -DBUILD_JPEG=y -DWITH_IPP=y -DENABLE_CONFIG_VERIFICATION=y -DBUILD_LIST=core,gapi,highgui,imgcodecs,imgproc,videoio ..
+          cmake -DCMAKE_BUILD_TYPE=Release -DWITH_INF_ENGINE=y -DOpenVINO_DIR=$GITHUB_WORKSPACE/ov/runtime/cmake/ -DCMAKE_CXX_COMPILER_LAUNCHER=ccache -DCMAKE_CXX_LINKER_LAUNCHER=ccache -DCMAKE_C_COMPILER_LAUNCHER=ccache -DCMAKE_C_LINKER_LAUNCHER=ccache -DBUILD_TESTS=y -DVIDEOIO_ENABLE_PLUGINS=y -DBUILD_PERF_TESTS=n -DBUILD_EXAMPLES=n -DBUILD_opencv_apps=y -DWITH_OPENCL=n -DWITH_OPENCLAMDBLAS=n -DWITH_GSTREAMER=n -DWITH_V4L=ON -DWITH_LIBV4L=ON -DWITH_OPENCLAMDFFT=n -DWITH_VA=n -DWITH_VA_INTEL=n -DWITH_PROTOBUF=n -DBUILD_PROTOBUF=n -DBUILD_JAVA=n -DBUILD_opencv_java_bindings_generator=n -DBUILD_opencv_python2=n -DBUILD_opencv_python3=n -DWITH_IMGCODEC_HDR=y -DWITH_IMGCODEC_SUNRASTER=y -DWITH_IMGCODEC_PXM=y -DWITH_IMGCODEC_PFM=y -DWITH_PNG=y -DWITH_TIFF=n -DWITH_WEBP=n -DWITH_OPENJPEG=n -DWITH_JASPER=n -DWITH_OPENEXR=n -DBUILD_opencv_dnn=n -DBUILD_opencv_features2d=n -DBUILD_opencv_flann=n -DWITH_TBB=n -DBUILD_INFO_SKIP_EXTRA_MODULES=n -DBUILD_JASPER=n -DBUILD_PNG=n -DBUILD_OPENEXR=n -DBUILD_WEBP=n -DBUILD_ZLIB=n -DWITH_CUDA=n -DWITH_EIGEN=n -DWITH_GPHOTO2=n -DOPENCV_GAPI_GSTREAMER=n -DWITH_LAPACK=n -DWITH_MATLAB=n -DWITH_MFX=n -DWITH_QUIRC=n -DWITH_VTK=n -DINSTALL_PDB=n -DINSTALL_TESTS=n -DINSTALL_C_EXAMPLES=n -DINSTALL_PYTHON_EXAMPLES=n -DOPENCV_GENERATE_SETUPVARS=n -DWITH_1394=n -DWITH_FFMPEG=y -DWITH_GTK_2_X=y -DBUILD_JPEG=y -DWITH_IPP=y -DENABLE_CONFIG_VERIFICATION=y -DBUILD_LIST=core,gapi,highgui,imgcodecs,imgproc,videoio,video ..
           cmake --build . -j $((`nproc`*2+2))
       - name: build_demos.sh
         run: |

diff --git a/demos/gaze_estimation_demo/cpp_gapi/main.cpp b/demos/gaze_estimation_demo/cpp_gapi/main.cpp
@@ -30,7 +30,7 @@
 #include <opencv2/gapi/gproto.hpp>
 #include <opencv2/gapi/gstreaming.hpp>
 #include <opencv2/gapi/infer.hpp>
-#include <opencv2/gapi/infer/ie.hpp>
+#include <opencv2/gapi/infer/ov.hpp>
 #include <opencv2/gapi/streaming/format.hpp>
 #include <opencv2/highgui.hpp>
 #include <opencv2/imgproc.hpp>
@@ -175,7 +175,7 @@ int main(int argc, char* argv[]) {
                                         processed_gaze_vectors));
         /** ---------------- End of graph ---------------- **/
         /** Configure networks **/
-        auto face_net = cv::gapi::ie::Params<nets::Faces>{
+        auto face_net = cv::gapi::ov::Params<nets::Faces>{
             FLAGS_m_fd,  // path to topology IR
             fileNameNoExt(FLAGS_m_fd) + ".bin",  // path to weights
             FLAGS_d_fd,  // device specifier
@@ -210,11 +210,11 @@ int main(int argc, char* argv[]) {
 
             if (std::fabs(imageAspectRatio - networkAspectRatio) > aspectRatioThreshold) {
                 inShape[3] = static_cast<unsigned long>(inShape[2] * imageAspectRatio);
-                face_net.cfgInputReshape(input.get_any_name(), inShape);
+                face_net.cfgReshape(inShape);
             }
         }
         auto head_net =
-            cv::gapi::ie::Params<nets::HeadPose>{
+            cv::gapi::ov::Params<nets::HeadPose>{
                 FLAGS_m_hp,  // path to topology IR
                 fileNameNoExt(FLAGS_m_hp) + ".bin",  // path to weights
                 FLAGS_d_hp,  // device specifier
@@ -223,7 +223,7 @@ int main(int argc, char* argv[]) {
         slog::info << "The Head Pose Estimation model " << FLAGS_m_hp << " is loaded to " << FLAGS_d_hp << " device."
                    << slog::endl;
 
-        auto landmarks_net = cv::gapi::ie::Params<nets::Landmarks>{
+        auto landmarks_net = cv::gapi::ov::Params<nets::Landmarks>{
             FLAGS_m_lm,  // path to topology IR
             fileNameNoExt(FLAGS_m_lm) + ".bin",  // path to weights
             FLAGS_d_lm,  // device specifier
@@ -233,7 +233,7 @@ int main(int argc, char* argv[]) {
 
         // clang-format off
         auto gaze_net =
-            cv::gapi::ie::Params<nets::Gaze>{
+            cv::gapi::ov::Params<nets::Gaze>{
                 FLAGS_m,  // path to topology IR
                 fileNameNoExt(FLAGS_m) + ".bin",  // path to weights
                 FLAGS_d,  // device specifier
@@ -242,7 +242,7 @@ int main(int argc, char* argv[]) {
         slog::info << "The Gaze Estimation model " << FLAGS_m << " is loaded to " << FLAGS_d << " device."
                    << slog::endl;
 
-        auto eyes_net = cv::gapi::ie::Params<nets::Eyes>{
+        auto eyes_net = cv::gapi::ov::Params<nets::Eyes>{
             FLAGS_m_es,  // path to topology IR
             fileNameNoExt(FLAGS_m_es) + ".bin",  // path to weights
             FLAGS_d_es,  // device specifier

diff --git a/demos/tests/cases.py b/demos/tests/cases.py
@@ -254,8 +254,7 @@ def single_option_cases(key, *args):
             device_keys=['-d', '-d_fd', '-d_hp', '-d_lm', '-d_es'],
             test_cases=combine_cases(
         TestCase(options={'-no_show': None,
-            **MONITORS,
-            '-i': TestDataArg('coco128/images/train2017/')}),
+            **MONITORS}),
         TestCase(options={
             '-m': ModelArg('gaze-estimation-adas-0002'),
             '-m_hp': ModelArg('head-pose-estimation-adas-0001'),
@@ -266,6 +265,10 @@ def single_option_cases(key, *args):
             '-m_fd',
             ModelArg('face-detection-adas-0001'),
             ModelArg('face-detection-retail-0004')),
+        single_option_cases(
+            '-i',
+            str('video.mp4'),
+            DataPatternArg('coco128-every-480x640x3')),
     )),
 
     CppDemo(name='gesture_recognition_demo', implementation='cpp_gapi',

diff --git a/demos/tests/run_tests.py b/demos/tests/run_tests.py
@@ -51,12 +51,14 @@
 from cases import Demo
 from data_sequences import DATA_SEQUENCES
 
+import requests
+
 scopes = {
     'base': importlib.import_module('cases').DEMOS,
     'performance': importlib.import_module('performance_cases').DEMOS,
 }
 COCO128_URL = "https://ultralytics.com/assets/coco128.zip"
-
+VIDEO_URL = "https://storage.openvinotoolkit.org/data/test_data/videos/head-pose-face-detection-male.mp4"
 
 def parser_paths_list(supported_devices):
     paths = supported_devices.split(',')
@@ -247,6 +249,10 @@ def main():
         with ZipFile(BytesIO(zipresp.read())) as zfile:
             zfile.extractall(args.test_data_dir)
 
+    r = requests.get(VIDEO_URL)
+    with open("video.mp4", 'wb') as f:
+        f.write(r.content)
+
     with temp_dir_as_path() as global_temp_dir:
         if args.models_dir:
             dl_dir = args.models_dir