Add face landmark detection support for Mediavision backend.
With this patch, move '_rects' member to each specific result structure,
and add _points to FldResultType for face landmark detection.
Change-Id: If0e801c6c01da247870e9ac3632d5953d90633c8
Signed-off-by: Inki Dae <inki.dae@samsung.com>
int bottom {};
};
+struct Point {
+ int x {};
+ int y {};
+};
+
using VecRect = std::vector<Rect>;
enum class DataType { NONE, FILE, IMAGE, RAW };
size_t size_in_bytes {};
};
-enum class ResultType { NONE, OBJECT_DETECTION, FACE_DETECTION, LANDMARK };
+enum class ResultType { NONE, OBJECT_DETECTION, FACE_DETECTION, FACE_LANDMARK };
struct BaseResultType {
ResultType _type { ResultType::NONE };
unsigned int _frame_number {};
- std::vector<Rect> _rects;
BaseResultType(ResultType type) : _type(type)
{}
virtual ~BaseResultType()
struct OdResultType : public BaseResultType {
OdResultType() : BaseResultType(ResultType::OBJECT_DETECTION)
{}
- // TODO
+ std::vector<Rect> _rects;
};
struct FdResultType : public BaseResultType {
FdResultType() : BaseResultType(ResultType::FACE_DETECTION)
{}
- // TODO
+ std::vector<Rect> _rects;
+};
+
+struct FldResultType : public BaseResultType {
+ FldResultType() : BaseResultType(ResultType::FACE_LANDMARK)
+ {}
+ std::vector<Point> _points;
};
enum class ServiceType { NONE, AUTO_ZOOM };
SET(SINGLEO_SERVICE_SOURCE_FILES
${SINGLEO_SERVICE_SOURCE_FILES}
${INFERENCE_MEDIAVISION_BACKEND_DIRECTORY}/src/MvFaceDetection.cpp
+ ${INFERENCE_MEDIAVISION_BACKEND_DIRECTORY}/src/MvFaceLandmark.cpp
${INFERENCE_MEDIAVISION_BACKEND_DIRECTORY}/src/MvObjectDetection.cpp
${INFERENCE_MEDIAVISION_BACKEND_DIRECTORY}/src/MvInferenceServiceFactory.cpp
)
-LIST(APPEND INFERENCE_LIBRARY_LIST ${INFERENCE_LIBRARY_LIST} mv_common mv_inference mv_object_detection)
+LIST(APPEND INFERENCE_LIBRARY_LIST ${INFERENCE_LIBRARY_LIST} mv_common mv_inference mv_object_detection mv_landmark_detection)
LIST(APPEND INFERENCE_HEADER_LIST ${INFERENCE_HEADER_LIST} ${INFERENCE_MEDIAVISION_BACKEND_DIRECTORY}/include /usr/include/media)
--- /dev/null
+/**
+ * Copyright (c) 2024 Samsung Electronics Co., Ltd All Rights Reserved
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef __MV_FACE_LANDMARK_H__
+#define __MV_FACE_LANDMARK_H__
+
+#include "IInferenceServiceInterface.h"
+#include "mv_facial_landmark_internal.h"
+#include "SingleoCommonTypes.h"
+
+namespace singleo
+{
+namespace inference
+{
+namespace backends
+{
+class MvFaceLandmark : public IInferenceServiceInterface
+{
+private:
+ mv_facial_landmark_h _handle {};
+ FldResultType _output_data;
+
+public:
+ MvFaceLandmark();
+ virtual ~MvFaceLandmark();
+
+ void configure() override;
+ void prepare() override;
+ void invoke(BaseDataType &input, bool async) override;
+ BaseResultType &result() override;
+};
+
+} // backends
+} // inference
+} // singleo
+
+#endif
--- /dev/null
+/**
+ * Copyright (c) 2024 Samsung Electronics Co., Ltd All Rights Reserved
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <stdexcept>
+#include "SingleoInputManager.h"
+#include "MvFaceLandmark.h"
+#include "SingleoLog.h"
+
+using namespace std;
+
+namespace singleo
+{
+namespace inference
+{
+namespace backends
+{
+MvFaceLandmark::MvFaceLandmark()
+{
+ int ret = mv_facial_landmark_create(&_handle);
+ if (ret != MEDIA_VISION_ERROR_NONE)
+ throw runtime_error("Fail to create face landmark detection handle.");
+}
+
+MvFaceLandmark::~MvFaceLandmark()
+{
+ mv_facial_landmark_destroy(_handle);
+}
+
+void MvFaceLandmark::configure()
+{
+ int ret = mv_facial_landmark_configure(_handle);
+ if (ret != MEDIA_VISION_ERROR_NONE)
+ throw runtime_error("Fail to configure face landmark detection.");
+}
+
+void MvFaceLandmark::prepare()
+{
+ int ret = mv_facial_landmark_prepare(_handle);
+ if (ret != MEDIA_VISION_ERROR_NONE)
+ throw runtime_error("Fail to prepare face landmark detection.");
+}
+
+void MvFaceLandmark::invoke(BaseDataType &input, bool async)
+{
+ ImageDataType &data = dynamic_cast<ImageDataType &>(input);
+
+ if (data._data_type != DataType::IMAGE) {
+ SINGLEO_LOGE("Invalid input type.");
+ throw invalid_argument("Input type not support.");
+ }
+
+ mv_source_h mv_src;
+
+ int ret = mv_create_source(&mv_src);
+ if (ret != MEDIA_VISION_ERROR_NONE)
+ throw runtime_error("Fail to create mv source.");
+
+ try {
+ ret = mv_source_fill_by_buffer(mv_src, data.ptr, data.width * data.height * data.byte_per_pixel, data.width,
+ data.height, MEDIA_VISION_COLORSPACE_RGB888);
+ if (ret != MEDIA_VISION_ERROR_NONE)
+ throw runtime_error("Fail to convert to mv source.");
+
+ ret = mv_facial_landmark_inference(_handle, mv_src);
+ if (ret != MEDIA_VISION_ERROR_NONE)
+ throw runtime_error("Fail to invoke face landmark detection.");
+ } catch (std::runtime_error &e) {
+ SINGLEO_LOGE("%s", e.what());
+ }
+
+ ret = mv_destroy_source(mv_src);
+ if (ret != MEDIA_VISION_ERROR_NONE)
+ throw runtime_error("Fail to destroy mv source.");
+}
+
+BaseResultType &MvFaceLandmark::result()
+{
+ unsigned long frame_number;
+ unsigned int result_cnt;
+
+ int ret = mv_facial_landmark_get_result_count(_handle, &frame_number, &result_cnt);
+ if (ret != MEDIA_VISION_ERROR_NONE)
+ throw runtime_error("Fail to get face landmark detection result count.");
+
+ _output_data._points.clear();
+ _output_data._frame_number = frame_number;
+
+ for (unsigned int idx = 0; idx < result_cnt; ++idx) {
+ Point point;
+
+ ret = mv_facial_landmark_get_position(_handle, idx, (unsigned int *) &point.x, (unsigned int *) &point.y);
+ if (ret != MEDIA_VISION_ERROR_NONE)
+ throw runtime_error("Fail to get face landmark detection point.");
+
+ _output_data._points.push_back(point);
+ }
+
+ return _output_data;
+}
+
+}
+}
+}
#include "InferenceServiceFactory.h"
#include "MvInferenceServiceFactory.h"
#include "MvFaceDetection.h"
+#include "MvFaceLandmark.h"
#include "MvObjectDetection.h"
#include "SingleoLog.h"
#include "SingleoException.h"
std::unique_ptr<IInferenceServiceInterface> MvInferenceServiceFactory::createFaceLandmarkDetection()
{
- throw InvalidOperation("Interface not supported yet.");
+ return make_unique<MvFaceLandmark>();
}
}
void AutoZoom::updateResult(BaseDataType &in_data)
{
auto &output_data = _inference_service->result();
- AutoZoomResult autozoom_result;
- vector<Rect> rects;
if (output_data._type != ResultType::OBJECT_DETECTION && output_data._type != ResultType::FACE_DETECTION)
throw InvalidParameter("Invalid result type");
- rects = output_data._rects;
+ vector<Rect> &rects = dynamic_cast<FdResultType &>(output_data)._rects;
+ AutoZoomResult autozoom_result;
+
autozoom_result.frame_number = output_data._frame_number;
autozoom_result.num_of_objects = rects.size();
projects / platform / core / api / singleo.git / commitdiff