add bindings for bounding box detection

README.md

@@ -14,22 +14,22 @@ This produces a shared library that can subsequently be used in CMake/Visual Stu
 - Bazel 5.2.0: https://bazel.build/install/ubuntu
 - Numpy: ```pip3 install numpy```
 
-#### Build and install OpenCV 4.7.0
+#### Build and install OpenCV 4.9.0
 ```
-wget -O opencv.zip https://github.com/opencv/opencv/archive/4.7.0.zip
-wget -O opencv_contrib.zip https://github.com/opencv/opencv_contrib/archive/4.7.0.zip
+wget -O opencv.zip https://github.com/opencv/opencv/archive/4.9.0.zip
+wget -O opencv_contrib.zip https://github.com/opencv/opencv_contrib/archive/4.9.0.zip
 unzip opencv.zip
 unzip opencv_contrib.zip
 mkdir -p opencv_build && cd opencv_build
-cmake -DOPENCV_EXTRA_MODULES_PATH=../opencv_contrib-4.7.0/modules ../opencv-4.7.0
+cmake -DOPENCV_EXTRA_MODULES_PATH=../opencv_contrib-4.9.0/modules ../opencv-4.9.0
 cmake --build .
 cmake --install .
 ```
 
 #### Run the build script
 ```
 cd <path to libmediapipe>
-./build-x86_64-linux.sh --version v0.8.11 --config release --opencv_dir /usr/local
+./build-x86_64-linux.sh --version v0.10.11 --config release --opencv_dir /usr/local
 ```
 
 ### Windows
@@ -43,12 +43,12 @@ cd <path to libmediapipe>
 - Git with Bash
 - Numpy: ```pip install numpy```
 
-#### Install OpenCV 4.7.0
-The installer can be downloaded from https://github.com/opencv/opencv/releases/download/4.7.0/opencv-4.7.0-windows.exe
+#### Install OpenCV 4.9.0
+The installer can be downloaded from https://github.com/opencv/opencv/releases/download/4.9.0/opencv-4.9.0-windows.exe
 
 #### Run the build script
 ```cd <path to libmediapipe>
-./build-x86_64-windows.ps1 --version v0.8.11 --config release --opencv_dir <path to opencv build>
+./build-x86_64-windows.ps1 --version v0.10.11 --config release --opencv_dir <path to opencv build>
 ```
 
 ### MacOS
@@ -59,14 +59,14 @@ The installer can be downloaded from https://github.com/opencv/opencv/releases/d
 - Xcode
 - Numpy: ```pip install numpy``` or ```pip3 install numpy```
 
-#### Build and install OpenCV 4.7.0
+#### Build and install OpenCV 4.9.0
 ```
-curl -L -o opencv.zip https://github.com/opencv/opencv/archive/4.7.0.zip
-curl -L -o opencv_contrib.zip https://github.com/opencv/opencv_contrib/archive/4.7.0.zip
+curl -L -o opencv.zip https://github.com/opencv/opencv/archive/4.9.0.zip
+curl -L -o opencv_contrib.zip https://github.com/opencv/opencv_contrib/archive/4.9.0.zip
 unzip opencv.zip
 unzip opencv_contrib.zip
 mkdir -p opencv_build && cd opencv_build
-cmake -DBUILD_ZLIB=false -DOPENCV_EXTRA_MODULES_PATH=../opencv_contrib-4.7.0/modules ../opencv-4.7.0
+cmake -DBUILD_ZLIB=false -DOPENCV_EXTRA_MODULES_PATH=../opencv_contrib-4.9.0/modules ../opencv-4.9.0
 cmake --build . -j6
 cmake --install .
 ```
@@ -79,7 +79,7 @@ cmake --install .
 
 ```
 cd <path to libmediapipe>
-./build-<arch>-macos.sh --version v0.8.11 --config release --opencv_dir /usr/local
+./build-<arch>-macos.sh --version v0.10.11 --config release --opencv_dir /usr/local
 ```
 
 ### Android
@@ -100,7 +100,7 @@ export ANDROID_NDK_HOME=<path to android ndk>
 #### Run the build script
 ```
 cd <path to libmediapipe>
-./build-aarch64-android.sh --version v0.8.11 --config release
+./build-aarch64-android.sh --version v0.10.11 --config release
 ```
 
 ## Usage

build-aarch64-android.sh

@@ -3,7 +3,7 @@
 set -e
 shopt -s nullglob
 
-VERSION="v0.8.11"
+VERSION="v0.10.11"
 OPENCV_DIR=""
 CONFIG="debug"
 

build-aarch64-macos.sh

@@ -3,7 +3,7 @@
 set -e
 shopt -s nullglob
 
-VERSION="v0.8.11"
+VERSION="v0.10.11"
 OPENCV_DIR=""
 CONFIG="debug"
 
@@ -149,7 +149,7 @@ echo "Done"
 
 echo -n "Patching dynamic library - "
 DYLIB_PATH=mediapipe/bazel-bin/mediapipe/c/libmediapipe.dylib
-mv mediapipe/bazel-bin/mediapipe/c/libmediapipe.so "$DYLIB_PATH"
+mv mediapipe/bazel-bin/mediapipe/c/libmediapipe.dylib "$DYLIB_PATH"
 chmod +w "$DYLIB_PATH"
 install_name_tool -id @rpath/libmediapipe.dylib "$DYLIB_PATH"
 echo "Done"

build-x86_64-linux.sh

@@ -3,7 +3,7 @@
 set -e
 shopt -s nullglob
 
-VERSION="v0.8.11"
+VERSION="v0.10.11"
 OPENCV_DIR=""
 CONFIG="debug"
 

build-x86_64-macos.sh

@@ -3,7 +3,7 @@
 set -e
 shopt -s nullglob
 
-VERSION="v0.8.11"
+VERSION="v0.10.11"
 OPENCV_DIR=""
 CONFIG="debug"
 

build-x86_64-windows.sh

@@ -1,7 +1,7 @@
 set -e
 shopt -s nullglob
 
-VERSION="v0.8.11"
+VERSION="v0.10.11"
 OPENCV_DIR=""
 CONFIG="debug"
 

c/mediapipe.cpp

@@ -3,6 +3,7 @@
 #include "mediapipe/framework/calculator_framework.h"
 #include "mediapipe/framework/port/parse_text_proto.h"
 #include "mediapipe/framework/formats/image_frame.h"
+#include "mediapipe/framework/formats/detection.pb.h"
 #include "mediapipe/framework/formats/landmark.pb.h"
 #include "mediapipe/framework/formats/rect.pb.h"
 #include "mediapipe/framework/tool/options_util.h"
@@ -59,6 +60,33 @@ struct mp_packet {
     mediapipe::Packet packet;
 };
 
+
+template<typename Detection>
+static mp_bounding_box_list* get_bounding_boxes(mp_packet* packet) {
+    const auto& detections = packet->packet.Get<std::vector<Detection>>();
+
+    auto* boxes_array = new mp_bounding_box[detections.size()];
+
+    for (int i = 0; i < detections.size(); i++) {
+        const Detection& detection = detections[i];
+        if (detection.location_data().format() == mediapipe::LocationData::RELATIVE_BOUNDING_BOX) {
+            const auto& bbox = detection.location_data().relative_bounding_box();
+            boxes_array[i] = {
+                detection.score_size() > 0 ? detection.score(0) : 0,
+                static_cast<float>(bbox.xmin()),
+                static_cast<float>(bbox.ymin()),
+                static_cast<float>(bbox.width()),
+                static_cast<float>(bbox.height())
+            };
+        }
+    }
+
+    return new mp_bounding_box_list {
+        boxes_array,
+        static_cast<int>(detections.size())
+    };
+}
+
 template<typename List, typename Landmark>
 static mp_multi_face_landmark_list* get_multi_face_landmarks(mp_packet* packet) {
     const auto& mp_data = packet->packet.template Get<std::vector<List>>();
@@ -355,6 +383,10 @@ MEDIAPIPE_API mp_multi_face_landmark_list* mp_get_multi_face_landmarks(mp_packet
     return get_multi_face_landmarks<mediapipe::LandmarkList, mediapipe::Landmark>(packet);
 }
 
+MEDIAPIPE_API mp_bounding_box_list* mp_get_bounding_boxes(mp_packet* packet) {
+    return get_bounding_boxes<mediapipe::Detection>(packet);
+}
+
 MEDIAPIPE_API mp_multi_face_landmark_list* mp_get_norm_multi_face_landmarks(mp_packet* packet) {
     return get_multi_face_landmarks<mediapipe::NormalizedLandmarkList, mediapipe::NormalizedLandmark>(packet);
 }
@@ -368,6 +400,11 @@ MEDIAPIPE_API void mp_destroy_multi_face_landmarks(mp_multi_face_landmark_list*
     delete multi_face_landmarks;
 }
 
+MEDIAPIPE_API void mp_destroy_bounding_boxes(mp_bounding_box_list* bounding_boxes) {
+    delete[] bounding_boxes->boxes;
+    delete bounding_boxes;
+}
+
 MEDIAPIPE_API mp_rect_list* mp_get_rects(mp_packet* packet) {
     return get_rects<mediapipe::Rect>(packet);
 }

c/mediapipe.h

@@ -100,6 +100,21 @@ typedef struct {
     int length;
 } mp_rect_list;
 
+struct mp_bounding_box {
+    float score;
+    float xmin;
+    float ymin;
+    float width;
+    float height;
+};
+
+// A list of bounding boxes
+struct mp_bounding_box_list {
+    mp_bounding_box* boxes;
+    int length;
+};
+
+
 /// Hand landmark indices in a mp_landmark_list.
 /// For more information, see: https://google.github.io/mediapipe/solutions/hands.html#hand-landmark-model
 typedef enum {
@@ -222,13 +237,20 @@ MEDIAPIPE_API void mp_copy_packet_image(mp_packet* packet, uint8_t* out_data);
 /// The list should be destroyed with mp_destroy_multi_face_landmarks.
 MEDIAPIPE_API mp_multi_face_landmark_list* mp_get_multi_face_landmarks(mp_packet* packet);
 
+/// Returns the multiface bounding box and score of a packet
+/// this list should be destroyed with mp_destroy_bounding_boxes
+MEDIAPIPE_API mp_bounding_box_list* mp_get_bounding_boxes(mp_packet* packet);
+
 /// Returns the normalized multi-face landmarks of a packet.
 /// The list should be destroyed with mp_destroy_multi_face_landmarks.
 MEDIAPIPE_API mp_multi_face_landmark_list* mp_get_norm_multi_face_landmarks(mp_packet* packet);
 
 /// Deallocates a multi-face landmark list.
 MEDIAPIPE_API void mp_destroy_multi_face_landmarks(mp_multi_face_landmark_list* multi_face_landmarks);
 
+// Dealocates a multiface bounding box list
+MEDIAPIPE_API void mp_destroy_bounding_boxes(mp_bounding_box_list* bounding_boxes);
+
 /// Returns the rectangles of a packet.
 /// The list should be destroyed with mp_destroy_rects.
 MEDIAPIPE_API mp_rect_list* mp_get_rects(mp_packet* packet);