const mv_point_s *locations,
void *user_data);
-typedef void (*mv_inference_pose_estimation_detected_cb)(
- mv_source_h source,
- int number_of_landmarks,
- const mv_point_s *locations,
- void *user_data);
-
/**
* @brief Performs facial landmarks detection on the @a source.
* @details Use this function to launch facial landmark detection.
void *user_data);
/**
- * @brief Performs pose estimation detection on the @a source.
- * @details Use this function to launch pose estimation detection.
- * Each time when mv_inference_pose_estimation_detect() is
- * called, @a detected_cb will receive a list pose estimation's locations
- * in the media source.
- *
- * @since_tizen 6.0
- * @remarks This function is synchronous and may take considerable time to run.
- *
- * @param[in] source The handle to the source of the media
- * @param[in] infer The handle to the inference
- * @param[in] roi Rectangular area including a face in @a source which
- * will be analyzed. If NULL, then the whole source will be
- * analyzed.
- * @param[in] detected_cb The callback which will receive the detection results.
- * @param[in] user_data The user data passed from the code where
- * mv_inference_pose_estimation_detect() is invoked.
- * This data will be accessible in @a detected_cb callback.
- * @return @c 0 on success, otherwise a negative error value
- * @retval #MEDIA_VISION_ERROR_NONE Successful
- * @retval #MEDIA_VISION_ERROR_NOT_SUPPORTED Not supported
- * @retval #MEDIA_VISION_ERROR_INVALID_PARAMETER Invalid parameter
- * @retval #MEDIA_VISION_ERROR_INTERNAL Internal error
- * @retval #MEDIA_VISION_ERROR_NOT_SUPPORTED_FORMAT Source colorspace
- * isn't supported
- *
- * @pre Create a source handle by calling mv_create_source()
- * @pre Create an inference handle by calling mv_inference_create()
- * @pre Configure an inference handle by calling mv_inference_configure()
- * @pre Prepare an inference by calling mv_inference_prepare()
- * @post @a detected_cb will be called to provide detection results
- *
- * @see mv_inference_pose_estimation_detected_cb()
- */
-int mv_inference_pose_estimation_detect(
- mv_source_h source,
- mv_inference_h infer,
- mv_rectangle_s *roi,
- mv_inference_pose_estimation_detected_cb detected_cb,
- void *user_data);
-
-/**
* @}
*/
std::vector<cv::Point> locations;
} FacialLandMarkDetectionResults; /**< structure FacialLandMarkDetectionResults */
-typedef struct _PoseEstimationResults {
- int number_of_pose_estimation;
- std::vector<cv::Point> locations;
-} PoseEstimationResults; /**< structure PoseEstimationResults */
-
namespace mediavision
{
namespace inference
*/
int GetFacialLandMarkDetectionResults(FacialLandMarkDetectionResults *results);
- /**
- * @brief Gets the PoseEstimationDetectionResults
- *
- * @since_tizen 6.0
- * @return @c true on success, otherwise a negative error value
- */
- int GetPoseEstimationDetectionResults(PoseEstimationResults *results);
-
int GetResults(std::vector<std::vector<int> > *dimInfo,
std::vector<float *> *results);
mv_inference_facial_landmark_detected_cb detected_cb,
void *user_data);
- /**
- * @brief Performs pose estimation detection on the @a source
- * @details Use this function to launch pose estimation detection.
- * Each time when mv_inference_pose_estimation_detect() is
- * called, @a detected_cb will receive a list pose estimation's locations
- * on the media source.
- *
- * @since_tizen 5.5
- *
- * @param [in] source The handle to the source of the media
- * @param [in] infer The handle to the inference
- * @param[in] roi Rectangular box bounding face image on the
- * @a source. If NULL, then full source will be
- * analyzed.
- * @param [in] detected_cb The callback which will be called for
- * detecting facial landmark on media source.
- * This callback will receive the detection results.
- * @param [in] user_data The user data passed from the code where
- * @ref mv_inference_facial_landmark_detect() is invoked.
- * This data will be accessible from @a detected_cb callback.
- *
- * @return @c 0 on success, otherwise a negative error value
- * @retval #MEDIA_VISION_ERROR_NONE Successful
- * @retval #MEDIA_VISION_ERROR_INVALID_PARAMETER Invalid parameter
- * @retval #MEDIA_VISION_ERROR_INTERNAL Internal error
- * @retval #MEDIA_VISION_ERROR_NOT_SUPPORTED_FORMAT Source colorspace
- * isn't supported
- * @retval #MEDIA_VISION_ERROR_OUT_OF_MEMORY Out of memory
- * @retval #MEDIA_VISION_ERROR_NOT_SUPPORTED Not supported
- *
- * @pre Create a source handle by calling @ref mv_create_source()
- * @pre Create an inference handle by calling @ref mv_inference_create()
- * @pre Configure an inference handle by calling @ref mv_inference_configure()
- * @pre Prepare an inference by calling @ref mv_inference_prepare()
- * @post @a detected_cb will be called to process detection results
- *
- * @see mv_inference_pose_estimation_detected_cb
- */
- int mv_inference_pose_estimation_detect_open(
- mv_source_h source, mv_inference_h infer, mv_rectangle_s *roi,
- mv_inference_pose_estimation_detected_cb detected_cb,
- void *user_data);
-
#ifdef __cplusplus
}
#endif /* __cplusplus */
return MEDIA_VISION_ERROR_NONE;
}
- int Inference::GetPoseEstimationDetectionResults(
- PoseEstimationResults *detectionResults)
- {
- tensor_t outputData;
-
- // Get inference result and contain it to outputData.
- int ret = FillOutputResult(outputData);
- if (ret != MEDIA_VISION_ERROR_NONE) {
- LOGE("Fail to get output result.");
- return ret;
- }
-
- std::vector<std::vector<int> > inferDimInfo(outputData.dimInfo);
- std::vector<void *> inferResults(outputData.data.begin(),
- outputData.data.end());
-
- long number_of_pose = inferDimInfo[0][3];
- float *tmp = static_cast<float *>(inferResults[0]);
- cv::Size heatMapSize(inferDimInfo[0][1], inferDimInfo[0][2]);
-
- cv::Point loc;
- double score;
- cv::Mat blurredHeatMap;
-
- cv::Mat reShapeTest(cv::Size(inferDimInfo[0][2], inferDimInfo[0][1]),
- CV_32FC(inferDimInfo[0][3]), (void *) tmp);
-
- cv::Mat multiChannels[inferDimInfo[0][3]];
- split(reShapeTest, multiChannels);
-
- float ratioX = static_cast<float>(mSourceSize.width) /
- static_cast<float>(inferDimInfo[0][2]);
- float ratioY = static_cast<float>(mSourceSize.height) /
- static_cast<float>(inferDimInfo[0][1]);
-
- PoseEstimationResults results;
- results.number_of_pose_estimation = 0;
- for (int poseIdx = 0; poseIdx < number_of_pose; poseIdx++) {
- cv::Mat heatMap = multiChannels[poseIdx];
-
- cv::GaussianBlur(heatMap, blurredHeatMap, cv::Size(), 5.0, 5.0);
- cv::minMaxLoc(heatMap, NULL, &score, NULL, &loc);
-
- LOGI("PoseIdx[%2d]: x[%2d], y[%2d], score[%.3f]", poseIdx, loc.x,
- loc.y, score);
- LOGI("PoseIdx[%2d]: x[%2d], y[%2d], score[%.3f]", poseIdx,
- static_cast<int>(static_cast<float>(loc.x + 1) * ratioX),
- static_cast<int>(static_cast<float>(loc.y + 1) * ratioY),
- score);
-
- loc.x = static_cast<int>(static_cast<float>(loc.x + 1) * ratioX);
- loc.y = static_cast<int>(static_cast<float>(loc.y + 1) * ratioY);
- results.locations.push_back(loc);
- results.number_of_pose_estimation++;
- }
-
- *detectionResults = results;
- LOGE("Inference: PoseEstimationResults: %d\n",
- results.number_of_pose_estimation);
- return MEDIA_VISION_ERROR_NONE;
- }
-
} /* Inference */
} /* MediaVision */
return ret;
#endif
-}
-
-int mv_inference_pose_estimation_detect(
- mv_source_h source, mv_inference_h infer, mv_rectangle_s *roi,
- mv_inference_pose_estimation_detected_cb detected_cb, void *user_data)
-{
- MEDIA_VISION_SUPPORT_CHECK(
- __mv_inference_face_check_system_info_feature_supported());
- MEDIA_VISION_INSTANCE_CHECK(source);
- MEDIA_VISION_INSTANCE_CHECK(infer);
- MEDIA_VISION_NULL_ARG_CHECK(detected_cb);
-
- MEDIA_VISION_FUNCTION_ENTER();
-
- int ret = MEDIA_VISION_ERROR_NONE;
-
-#ifdef MEDIA_VISION_INFERENCE_LICENCE_PORT
- /*
- ret = mv_inference_pose_estimation_lic(source, infer, detected_cb, user_data);
- */
-#else
-
- ret = mv_inference_pose_estimation_detect_open(source, infer, roi,
- detected_cb, user_data);
-
- MEDIA_VISION_FUNCTION_LEAVE();
-
- return ret;
-#endif
}
\ No newline at end of file
detected_cb(source, numberOfLandmarks, locations.data(), user_data);
return ret;
-}
-
-int mv_inference_pose_estimation_detect_open(
- mv_source_h source, mv_inference_h infer, mv_rectangle_s *roi,
- mv_inference_pose_estimation_detected_cb detected_cb, void *user_data)
-{
- Inference *pInfer = static_cast<Inference *>(infer);
-
- int ret = MEDIA_VISION_ERROR_NONE;
- int numberOfPoseEstimation = 0;
- std::vector<mv_source_h> sources;
- std::vector<mv_rectangle_s> rects;
-
- sources.push_back(source);
-
- if (roi != NULL)
- rects.push_back(*roi);
-
- ret = pInfer->Run(sources, rects);
- if (ret != MEDIA_VISION_ERROR_NONE) {
- LOGE("Fail to run inference");
- return ret;
- }
-
- PoseEstimationResults poseEstimationResults;
- ret = pInfer->GetPoseEstimationDetectionResults(&poseEstimationResults);
- if (ret != MEDIA_VISION_ERROR_NONE) {
- LOGE("Fail to get inference results");
- return ret;
- }
-
- numberOfPoseEstimation = poseEstimationResults.number_of_pose_estimation;
-
- std::vector<mv_point_s> locations(numberOfPoseEstimation);
-
- for (int n = 0; n < numberOfPoseEstimation; ++n) {
- locations[n].x = poseEstimationResults.locations[n].x;
- locations[n].y = poseEstimationResults.locations[n].y;
- }
-
- detected_cb(source, numberOfPoseEstimation, locations.data(), user_data);
-
- return ret;
-}
+}
\ No newline at end of file
#define FLD_OPENCV_CONFIG_CAFFE_PATH \
"/usr/share/capi-media-vision/models/FLD/caffe/fld_caffe_model_tweak.prototxt"
-//Pose Estimation
-#define PE_TFLITE_WEIGHT_PATH \
- "/usr/share/capi-media-vision/models/PE/tflite/ped_tflite_model.tflite"
-
/******
* Public model:
* IC: mobilenet caffe, tf?
return MEDIA_VISION_ERROR_NONE;
}
-int perform_armnn_pose_estimation_detection(mv_engine_config_h *engine_cfg)
-{
- int err = MEDIA_VISION_ERROR_NONE;
-
- mv_engine_config_h handle = NULL;
- err = mv_create_engine_config(&handle);
- if (err != MEDIA_VISION_ERROR_NONE) {
- printf("Fail to create engine configuration handle.\n");
- if (handle) {
- int err2 = mv_destroy_engine_config(handle);
- if (err2 != MEDIA_VISION_ERROR_NONE) {
- printf("Fail to destroy engine cofniguration.\n");
- }
- }
- return err;
- }
-
- char *inputNodeName = "image";
- char *outputNodeName[] = { "Convolutional_Pose_Machine/stage_5_out" };
-
- mv_engine_config_set_string_attribute(
- handle, MV_INFERENCE_MODEL_WEIGHT_FILE_PATH, PE_TFLITE_WEIGHT_PATH);
-
- mv_engine_config_set_int_attribute(handle, MV_INFERENCE_INPUT_DATA_TYPE,
- MV_INFERENCE_DATA_FLOAT32);
-
- mv_engine_config_set_double_attribute(handle, MV_INFERENCE_MODEL_MEAN_VALUE,
- 0.0);
-
- mv_engine_config_set_double_attribute(handle, MV_INFERENCE_MODEL_STD_VALUE,
- 1.0);
-
- mv_engine_config_set_int_attribute(handle, MV_INFERENCE_BACKEND_TYPE,
- MV_INFERENCE_BACKEND_ARMNN);
-
- mv_engine_config_set_int_attribute(handle, MV_INFERENCE_TARGET_TYPE,
- MV_INFERENCE_TARGET_CPU);
-
- mv_engine_config_set_int_attribute(handle, MV_INFERENCE_INPUT_TENSOR_WIDTH,
- 192);
-
- mv_engine_config_set_int_attribute(handle, MV_INFERENCE_INPUT_TENSOR_HEIGHT,
- 192);
-
- mv_engine_config_set_int_attribute(handle,
- MV_INFERENCE_INPUT_TENSOR_CHANNELS, 3);
-
- mv_engine_config_set_string_attribute(handle, MV_INFERENCE_INPUT_NODE_NAME,
- inputNodeName);
-
- mv_engine_config_set_array_string_attribute(
- handle, MV_INFERENCE_OUTPUT_NODE_NAMES, outputNodeName, 1);
-
- *engine_cfg = handle;
- return err;
-}
-
-int perform_pose_estimation_detection()
-{
- int err = MEDIA_VISION_ERROR_NONE;
-
- int sel_opt = 0;
- const int options[] = { 1, 2, 3, 4, 5 };
- const *names[] = { "Configuration", "ARMNN(CPU) + PoseEstimation",
- "Prepare", "Run", "Back" };
-
- mv_engine_config_h engine_cfg = NULL;
- mv_inference_h infer = NULL;
- mv_source_h mvSource = NULL;
-
- while (sel_opt == 0) {
- sel_opt = show_menu("Select Action:", options, names,
- ARRAY_SIZE(options));
- switch (sel_opt) {
- case 1: {
- //perform configuration
- if (engine_cfg) {
- int err2 = mv_destroy_engine_config(engine_cfg);
- if (err2 != MEDIA_VISION_ERROR_NONE)
- printf("Fail to destroy engine_cfg [err:%i]\n", err2);
- }
-
- err = perform_configuration(&engine_cfg);
- } break;
- case 2: {
- //perform pose estimation config
- if (engine_cfg) {
- int err2 = mv_destroy_engine_config(engine_cfg);
- if (err2 != MEDIA_VISION_ERROR_NONE)
- printf("Fail to destroy engine_cfg [err:%i]\n", err2);
- }
- err = perform_armnn_pose_estimation_detection(&engine_cfg);
- } break;
- case 3: {
- // create - configure - prepare
- if (infer) {
- int err2 = mv_inference_destroy(infer);
- if (err2 != MEDIA_VISION_ERROR_NONE) {
- printf("Fail to destroy inference handle [err:%i]\n", err2);
- }
- }
-
- // inference
- // create handle
- err = mv_inference_create(&infer);
- if (err != MEDIA_VISION_ERROR_NONE) {
- printf("Fail to create inference handle [err:%i]\n", err);
- break;
- }
-
- //configure
- err = mv_inference_configure(infer, engine_cfg);
- if (err != MEDIA_VISION_ERROR_NONE) {
- printf("Fail to configure inference handle [err:%i]\n", err);
- break;
- }
-
- //prepare
- err = mv_inference_prepare(infer);
- if (err != MEDIA_VISION_ERROR_NONE) {
- printf("Fail to prepare inference handle");
- break;
- }
- } break;
- case 4: {
- if (mvSource) {
- int err2 = mv_destroy_source(mvSource);
- if (err2 != MEDIA_VISION_ERROR_NONE)
- ;
- printf("Fail to destroy mvSource\n");
- }
-
- char *in_file_name = NULL;
- /* Load media source */
- while (input_string("Input file name to be inferred:", 1024,
- &(in_file_name)) == -1)
- printf("Incorrect input! Try again.\n");
-
- err = mv_create_source(&mvSource);
- if (err != MEDIA_VISION_ERROR_NONE) {
- printf("Fail to create mvSource.\n");
- free(in_file_name);
- break;
- }
-
- err = load_mv_source_from_file(in_file_name, mvSource);
- if (err != MEDIA_VISION_ERROR_NONE) {
- int err2 = mv_destroy_source(mvSource);
- if (err2 != MEDIA_VISION_ERROR_NONE) {
- printf("Fail to destroy mvSource.\n", err2);
- }
- free(in_file_name);
- break;
- }
- free(in_file_name);
-
- struct timespec s_tspec;
- struct timespec e_tspec;
-
- clock_gettime(CLOCK_MONOTONIC, &s_tspec);
-
- // Pose estimation
- err = mv_inference_pose_estimation_detect(
- mvSource, infer, NULL, _pose_estimation_detected_cb, NULL);
-
- clock_gettime(CLOCK_MONOTONIC, &e_tspec);
-
- struct timespec diffspec = diff(s_tspec, e_tspec);
- unsigned long timeDiff = gettotalmillisec(diffspec);
- printf("elased time : %lu(ms)\n", timeDiff);
-
- break;
- }
- case 5: {
- //perform destroy
- if (engine_cfg) {
- err = mv_destroy_engine_config(engine_cfg);
- if (err != MEDIA_VISION_ERROR_NONE) {
- printf("Fail to destroy engine_cfg [err:%i]\n", err);
- }
- }
-
- if (infer) {
- err = mv_inference_destroy(infer);
- if (err != MEDIA_VISION_ERROR_NONE) {
- printf("Fail to destroy inference handle [err:%i]\n", err);
- }
- }
- } break;
- default:
- printf("Invalid option.\n");
- sel_opt = 0;
- continue;
- }
-
- int do_another = 0;
- if (err != MEDIA_VISION_ERROR_NONE) {
- printf("ERROR: Action is finished with error code: %i\n");
- }
-
- sel_opt = 0;
- const int options_last[] = { 1, 2 };
- const char *names_last[] = { "Yes", "No" };
-
- while (sel_opt == 0) {
- sel_opt = show_menu(
- "Run Pose Estimation Detection again?:", options_last,
- names_last, ARRAY_SIZE(options_last));
- switch (sel_opt) {
- case 1:
- do_another = 1;
- break;
- case 2:
- do_another = 0;
- break;
- default:
- printf("Invalid option.\n");
- sel_opt = 0;
- }
- }
-
- sel_opt = (do_another == 1) ? 0 : 1;
- }
-
- return MEDIA_VISION_ERROR_NONE;
-}
-
int main()
{
int sel_opt = 0;
- const int options[] = { 1, 2, 3, 4, 5, 6 };
+ const int options[] = { 1, 2, 3, 4, 5 };
const char *names[] = { "Image Classification", "Object Detection",
"Face Detection", "Facial LandmarkDetection",
- "Pose Estimation", "Exit" };
+ "Exit" };
int err = MEDIA_VISION_ERROR_NONE;
while (sel_opt == 0) {
break;
}
case 5: {
- err = perform_pose_estimation_detection();
- if (err != MEDIA_VISION_ERROR_NONE) {
- printf("Fail to perform pose estimation");
- }
- break;
- }
- case 6: {
printf("Exit");
} break;
default:
projects / platform / core / api / mediavision.git / commitdiff