add modelAPI to pedestian_tracker_demo

demos/common/cpp/models/src/detection_model_yolo.cpp

@@ -155,32 +155,32 @@ void ModelYolo::prepareInputsOutputs(InferenceEngine::CNNNetwork& cnnNetwork) {
         }
 
         std::sort(outputsNames.begin(), outputsNames.end(),
-            [&outputInfo](const std::string& x, const std::string&  y) {return outputInfo[x]->getDims()[2] > outputInfo[y]->getDims()[2];});
+            [&outputInfo](const std::string& x, const std::string& y) {return outputInfo[x]->getDims()[2] > outputInfo[y]->getDims()[2];});
 
         for (const auto& name : outputsNames) {
             auto& output = outputInfo[name];
             auto shape = output->getDims();
-            auto classes = shape[1] / num - 4 - isObjConf;
+            int classes = (int)shape[1] / num - 4 - (isObjConf ? 1 : 0);
             if (shape[1] % num != 0) {
                 throw std::runtime_error(std::string("The output blob ") + name + " has wrong 2nd dimension");
             }
             regions.emplace(name, Region(classes, 4,
                 presetAnchors.size() ? presetAnchors : defaultAnchors[yoloVersion],
                 std::vector<int64_t>(chosenMasks.begin() + i*num, chosenMasks.begin() + (i+1)*num),
-                shape[3], shape[2]));
+                (int)shape[3], (int)shape[2]));
             i++;
         }
     }
     else {
         // Currently externally set anchors and masks are supported only for YoloV4
-        if(presetAnchors.size() || presetMasks.size()){
+        if(presetAnchors.size() || presetMasks.size()) {
             slog::warn << "Preset anchors and mask can be set for YoloV4 model only. "
                 "This model is not YoloV4, so these options will be ignored." << slog::endl;
         }
     }
 }
 
-std::unique_ptr<ResultBase> ModelYolo::postprocess(InferenceResult & infResult) {
+std::unique_ptr<ResultBase> ModelYolo::postprocess(InferenceResult& infResult) {
     DetectionResult* result = new DetectionResult(infResult.frameId, infResult.metaData);
     std::vector<DetectedObject> objects;
 
@@ -333,8 +333,8 @@ ModelYolo::Region::Region(const std::shared_ptr<ngraph::op::RegionYolo>& regionY
     num = mask.size();
 
     auto shape = regionYolo->get_input_shape(0);
-    outputWidth = shape[3];
-    outputHeight = shape[2];
+    outputWidth = (int)shape[3];
+    outputHeight = (int)shape[2];
 
     if (num) {
 

demos/pedestrian_tracker_demo/cpp/CMakeLists.txt

@@ -1,12 +1,15 @@
-# Copyright (C) 2018-2019 Intel Corporation
+# Copyright (C) 2018-2021 Intel Corporation
 # SPDX-License-Identifier: Apache-2.0
 #
 
-file(GLOB_RECURSE SOURCES ${CMAKE_CURRENT_SOURCE_DIR}/*.cpp)
-file(GLOB_RECURSE HEADERS ${CMAKE_CURRENT_SOURCE_DIR}/*.hpp)
+file(GLOB_RECURSE SOURCES ${CMAKE_CURRENT_SOURCE_DIR}/*.cpp ../${CMAKE_CURRENT_SOURCE_DIR}/*.cpp)
+file(GLOB_RECURSE HEADERS ${CMAKE_CURRENT_SOURCE_DIR}/*.hpp ../${CMAKE_CURRENT_SOURCE_DIR}/*.hpp)
+
+
+
 
 add_demo(NAME pedestrian_tracker_demo
     SOURCES ${SOURCES}
     HEADERS ${HEADERS}
     INCLUDE_DIRECTORIES "${CMAKE_CURRENT_SOURCE_DIR}/include"
-    DEPENDENCIES monitors)
+    DEPENDENCIES monitors models pipelines)

demos/pedestrian_tracker_demo/cpp/README.md

@@ -40,12 +40,58 @@ python3 <omz_dir>/tools/downloader/converter.py --list models.lst
 
 ### Supported Models
 
-* person-detection-retail-0002
-* person-detection-retail-0013
-* person-reidentification-retail-0277
-* person-reidentification-retail-0286
-* person-reidentification-retail-0287
-* person-reidentification-retail-0288
+* architecture_type = centernet
+  - ctdet_coco_dlav0_384
+  - ctdet_coco_dlav0_512
+* architecture_type = ssd
+  - efficientdet-d0-tf
+  - efficientdet-d1-tf
+  - faster-rcnn-resnet101-coco-sparse-60-0001
+  - pedestrian-and-vehicle-detector-adas-0001
+  - pedestrian-detection-adas-0002
+  - pelee-coco
+  - person-detection-0106
+  - person-detection-0200
+  - person-detection-0201
+  - person-detection-0202
+  - person-detection-0203
+  - person-detection-retail-0002
+  - person-detection-retail-0013
+  - person-vehicle-bike-detection-2000
+  - person-vehicle-bike-detection-2001
+  - person-vehicle-bike-detection-2002
+  - person-vehicle-bike-detection-2003
+  - person-vehicle-bike-detection-2004
+  - rfcn-resnet101-coco-tf
+  - retinanet-tf
+  - ssd300
+  - ssd512
+  - ssd-resnet34-1200-onnx
+  - ssd_mobilenet_v1_coco
+  - ssd_mobilenet_v1_fpn_coco
+  - ssd_mobilenet_v2_coco
+  - ssd_resnet50_v1_fpn_coco
+  - ssdlite_mobilenet_v2
+  - vehicle-detection-adas-0002
+* architecture_type = yolo
+  - person-vehicle-bike-detection-crossroad-yolov3-1020
+  - yolo-v3-tf
+  - yolo-v3-tiny-tf
+  - yolo-v1-tiny-tf
+  - yolo-v2-ava-0001
+  - yolo-v2-ava-sparse-35-0001
+  - yolo-v2-ava-sparse-70-0001
+  - yolo-v2-tf
+  - yolo-v2-tiny-ava-0001
+  - yolo-v2-tiny-ava-sparse-30-0001
+  - yolo-v2-tiny-ava-sparse-60-0001
+  - yolo-v2-tiny-tf
+  - yolo-v2-tiny-vehicle-detection-0001
+* reidentification models
+  - person-reidentification-retail-0277
+  - person-reidentification-retail-0286
+  - person-reidentification-retail-0287
+  - person-reidentification-retail-0288
 
 > **NOTE**: Refer to the tables [Intel's Pre-Trained Models Device Support](../../../models/intel/device_support.md) and [Public Pre-Trained Models Device Support](../../../models/public/device_support.md) for the details on models inference support at different devices.
 
@@ -69,6 +115,7 @@ Options:
     -o "<path>"                  Optional. Name of the output file(s) to save.
     -limit "<num>"               Optional. Number of frames to store in output. If 0 is set, all frames are stored.
     -m_det "<path>"              Required. Path to the Pedestrian Detection Retail model (.xml) file.
+    -at \"<type>\"                  Required. Architecture type for detector model: centernet, ssd or yolo.
     -m_reid "<path>"             Required. Path to the Pedestrian Reidentification Retail model (.xml) file.
     -l "<absolute_path>"         Optional. For CPU custom layers, if any. Absolute path to a shared library with the kernels implementation.
           Or
@@ -80,6 +127,15 @@ Options:
     -delay                       Optional. Delay between frames used for visualization. If negative, the visualization is turned off (like with the option 'no_show'). If zero, the visualization is made frame-by-frame.
     -out "<path>"                Optional. The file name to write output log file with results of pedestrian tracking. The format of the log file is compatible with MOTChallenge format.
     -u                           Optional. List of monitors to show initially.
+	-t                           Optional. Probability threshold for detections.
+    -auto_resize                 Optional. Enables resizable input with support of ROI crop & auto resize.
+    -iou_t                       Optional. Filtering intersection over union threshold for overlapping boxes.
+    -yolo_af                     Optional. Use advanced postprocessing/filtering algorithm for YOLO.
+    -labels \"<path>\"           Optional. Path to a file with labels mapping.
+    -nireq \"<integer>\"         Optional. Number of infer requests for detector model. If this option is omitted, number of infer requests is determined automatically.
+    -nstreams                    Optional. Number of streams to use for inference on the CPU or/and GPU in throughput mode for detector model (for HETERO and MULTI device cases use format <device1>:<nstreams1>,<device2>:<nstreams2> or just <nstreams>)
+    -nthreads \"<integer>\"      Optional. Number of threads for detector model.
+    -person_label                Optional. Label of class person for detector. Default -1 for tracking all objects
 ```
 
 For example, to run the application with the OpenVINO&trade; toolkit pre-trained models with inferencing pedestrian detector on a GPU and pedestrian reidentification on a CPU, run the following command:
@@ -89,6 +145,7 @@ For example, to run the application with the OpenVINO&trade; toolkit pre-trained
                           -m_det <path_to_model>/person-detection-retail-0013.xml \
                           -m_reid <path_to_model>/person-reidentification-retail-0277.xml \
                           -d_det GPU
+                          -at ssd
 ```
 
 >**NOTE**: If you provide a single image as an input, the demo processes and renders it quickly, then exits. To continuously visualize inference results on the screen, apply the `loop` option, which enforces processing a single image in a loop.

demos/pedestrian_tracker_demo/cpp/include/cnn.hpp

@@ -1,4 +1,4 @@
-// Copyright (C) 2018-2019 Intel Corporation
+// Copyright (C) 2018-2021 Intel Corporation
 // SPDX-License-Identifier: Apache-2.0
 //
 
@@ -17,8 +17,8 @@
 /**
  * @brief Base class of config for network
  */
-struct CnnConfig {
-    explicit CnnConfig(const std::string& path_to_model)
+struct CnnConfigTracker {
+    explicit CnnConfigTracker(const std::string& path_to_model)
         : path_to_model(path_to_model) {}
 
     /** @brief Path to model description */
@@ -32,7 +32,7 @@ struct CnnConfig {
  */
 class CnnBase {
 public:
-    using Config = CnnConfig;
+    using Config = CnnConfigTracker;
 
     /**
      * @brief Constructor
@@ -92,7 +92,7 @@ class CnnBase {
 
 class VectorCNN : public CnnBase {
 public:
-    VectorCNN(const CnnConfig& config,
+    VectorCNN(const CnnConfigTracker& config,
               const InferenceEngine::Core & ie,
               const std::string & deviceName);
 

demos/pedestrian_tracker_demo/cpp/include/core.hpp

@@ -1,4 +1,4 @@
-// Copyright (C) 2018-2019 Intel Corporation
+// Copyright (C) 2018-2021 Intel Corporation
 // SPDX-License-Identifier: Apache-2.0
 //
 
@@ -17,7 +17,7 @@
 struct TrackedObject {
     cv::Rect rect;       ///< Detected object ROI (zero area if N/A).
     double confidence;   ///< Detection confidence level (-1 if N/A).
-    int frame_idx;       ///< Frame index where object was detected (-1 if N/A).
+    int64_t frame_idx;       ///< Frame index where object was detected (-1 if N/A).
     int object_id;       ///< Unique object identifier (-1 if N/A).
     uint64_t timestamp;  ///< Timestamp in milliseconds.
 
@@ -37,7 +37,7 @@ struct TrackedObject {
     /// \param frame_idx Index of frame.
     /// \param object_id Object ID.
     ///
-    TrackedObject(const cv::Rect &rect, float confidence, int frame_idx,
+    TrackedObject(const cv::Rect &rect, float confidence, int64_t frame_idx,
                   int object_id)
         : rect(rect),
         confidence(confidence),

demos/pedestrian_tracker_demo/cpp/include/descriptor.hpp

@@ -1,4 +1,4 @@
-// Copyright (C) 2018-2019 Intel Corporation
+// Copyright (C) 2018-2021 Intel Corporation
 // SPDX-License-Identifier: Apache-2.0
 //
 
@@ -107,9 +107,9 @@ class DescriptorIE : public IImageDescriptor {
     VectorCNN handler;
 
 public:
-    DescriptorIE(const CnnConfig& config,
+    DescriptorIE(const CnnConfigTracker& config,
                  const InferenceEngine::Core& ie,
-                 const std::string & deviceName):
+                 const std::string& deviceName):
         handler(config, ie, deviceName) {}
 
     ///

demos/pedestrian_tracker_demo/cpp/include/pedestrian_tracker_demo.hpp

@@ -1,4 +1,4 @@
-// Copyright (C) 2018-2019 Intel Corporation
+// Copyright (C) 2018-2021 Intel Corporation
 // SPDX-License-Identifier: Apache-2.0
 //
 
@@ -40,6 +40,18 @@ static const char delay_message[] = "Optional. Delay between frames used for vis
 static const char output_log_message[] = "Optional. The file name to write output log file with results of pedestrian tracking. "
                                           "The format of the log file is compatible with MOTChallenge format.";
 static const char utilization_monitors_message[] = "Optional. List of monitors to show initially.";
+static const char at_message[] = "Required. Architecture type for detector model: centernet, ssd or yolo.";
+static const char thresh_output_message[] = "Optional. Probability threshold for detections.";
+static const char input_resizable_message[] = "Optional. Enables resizable input with support of ROI crop & auto resize.";
+static const char iou_thresh_output_message[] = "Optional. Filtering intersection over union threshold for overlapping boxes.";
+static const char yolo_af_message[] = "Optional. Use advanced postprocessing/filtering algorithm for YOLO.";
+static const char labels_message[] = "Optional. Path to a file with labels mapping.";
+static const char nireq_message[] = "Optional. Number of infer requests for detector model. If this option is omitted, number of infer requests is determined automatically.";
+static const char num_threads_message[] = "Optional. Number of threads for detector model.";
+static const char num_streams_message[] = "Optional. Number of streams to use for inference on the CPU or/and GPU in "
+"throughput mode for detector model (for HETERO and MULTI device cases use format "
+"<device1>:<nstreams1>,<device2>:<nstreams2> or just <nstreams>)";
+static const char person_label_message[] = "Optional. Label of class person for detector. Default -1 for tracking all objects";
 
 
 DEFINE_bool(h, false, help_message);
@@ -56,6 +68,16 @@ DEFINE_bool(no_show, false, no_show_message);
 DEFINE_int32(delay, 3, delay_message);
 DEFINE_string(out, "", output_log_message);
 DEFINE_string(u, "", utilization_monitors_message);
+DEFINE_string(at, "", at_message);
+DEFINE_double(t, 0.5, thresh_output_message);
+DEFINE_bool(auto_resize, false, input_resizable_message);
+DEFINE_double(iou_t, 0.5, iou_thresh_output_message);
+DEFINE_bool(yolo_af, true, yolo_af_message);
+DEFINE_string(labels, "", labels_message);
+DEFINE_uint32(nireq, 0, nireq_message);
+DEFINE_uint32(nthreads, 0, num_threads_message);
+DEFINE_string(nstreams, "", num_streams_message);
+DEFINE_int32(person_label, -1, person_label_message);
 
 
 /**
@@ -85,4 +107,14 @@ static void showUsage() {
     std::cout << "    -delay                       " << delay_message << std::endl;
     std::cout << "    -out \"<path>\"                " << output_log_message << std::endl;
     std::cout << "    -u                           " << utilization_monitors_message << std::endl;
+    std::cout << "    -at \"<type>\"              " << at_message << std::endl;
+    std::cout << "    -t                          " << thresh_output_message << std::endl;
+    std::cout << "    -auto_resize                " << input_resizable_message << std::endl;
+    std::cout << "    -iou_t                      " << iou_thresh_output_message << std::endl;
+    std::cout << "    -yolo_af                    " << yolo_af_message << std::endl;
+    std::cout << "    -labels \"<path>\"          " << labels_message << std::endl;
+    std::cout << "    -nireq \"<integer>\"        " << nireq_message << std::endl;
+    std::cout << "    -nstreams                   " << num_streams_message << std::endl;
+    std::cout << "    -nthreads \"<integer>\"     " << num_threads_message << std::endl;
+    std::cout << "    -person_label               " << person_label_message << std::endl;
 }

demos/pedestrian_tracker_demo/cpp/include/tracker.hpp

@@ -1,4 +1,4 @@
-// Copyright (C) 2018-2019 Intel Corporation
+// Copyright (C) 2018-2021 Intel Corporation
 // SPDX-License-Identifier: Apache-2.0
 //
 
@@ -303,8 +303,8 @@ class PedestrianTracker {
 
 private:
     struct Match {
-        int frame_idx1;
-        int frame_idx2;
+        int64_t frame_idx1;
+        int64_t frame_idx2;
         cv::Rect rect1;
         cv::Rect rect2;
         cv::Rect pr_rect1;

demos/pedestrian_tracker_demo/cpp/include/utils.hpp

@@ -1,4 +1,4 @@
-// Copyright (C) 2018-2019 Intel Corporation
+// Copyright (C) 2018-2021 Intel Corporation
 // SPDX-License-Identifier: Apache-2.0
 //
 
@@ -24,7 +24,7 @@
 ///
 struct DetectionLogEntry {
     TrackedObjects objects;  ///< Detected objects.
-    int frame_idx;           ///< Processed frame index (-1 if N/A).
+    int64_t frame_idx;           ///< Processed frame index (-1 if N/A).
     double time_ms;          ///< Frame processing time in ms (-1 if N/A).
 
     ///