add label_ids output in crop node (#986)

src/custom_nodes/model_zoo_intel_object_detection/README.md

@@ -48,8 +48,9 @@ make BASE_OS=redhat
 | Output name        | Description           | Shape  | Precision |
 | ------------- |:-------------:| -----:| -----:|
 | images      | Returns images representing detected text boxes. Boxes are filtered based on confidence_threshold param. Resolution is defined by the node parameters. All images are in a single batch. Batch size depend on the number of detected objects.  | `N,1,C,H,W` | FP32 |
-| coordinates      | For every detected box `N` the following info is added: x coordinate for the box center, y coordinate for the box center, box original width, box original height | `N,1,4` | I32 |
+| coordinates      | For every detected box `N` the following info is added: x coordinate for the box center, y coordinate for the box center, box original width, box original height | `N,1,4` | FP32 |
 | confidences |   For every detected box `N` information about confidence level (N - number of detected boxes; more about demultiplexing [here](./../../../docs/demultiplexing.md)) | `N,1,1` | FP32 |
+| label_ids   |   For every detected box `N` information about label id (N - number of detected boxes; more about demultiplexing [here](./../../../docs/demultiplexing.md)) | `N,1,1` | I32 |
 
 # Custom node parameters
 Parameters can be defined in pipeline definition in OVMS configuration file. [Read more](./../../../docs/custom_node_development.md) about node parameters.

src/custom_nodes/model_zoo_intel_object_detection/model_zoo_intel_object_detection.cpp

@@ -44,6 +44,7 @@ static constexpr const char* OUTPUT_TENSOR_INFO_NAME = "output_info";
 static constexpr const char* OUTPUT_COORDINATES_INFO_DIMS_NAME = "coordinates_info_dims";
 static constexpr const char* OUTPUT_IMAGES_INFO_DIMS_NAME = "images_info_dims";
 static constexpr const char* OUTPUT_CONFIDENCES_INFO_DIMS_NAME = "confidences_info_dims";
+static constexpr const char* OUTPUT_LABEL_IDS_TENSOR_NAME = "label_ids";
 
 static constexpr const int QUEUE_SIZE = 1;
 
@@ -170,6 +171,26 @@ bool copy_confidences_into_output(struct CustomNodeTensor* output, const std::ve
     return true;
 }
 
+bool copy_label_ids_into_output(struct CustomNodeTensor* output, const std::vector<int>& labelIds) {
+    const uint64_t outputBatch = labelIds.size();
+    uint64_t byteSize = sizeof(float) * outputBatch;
+
+    float* buffer = (float*)malloc(byteSize);
+    NODE_ASSERT(buffer != nullptr, "malloc has failed");
+    std::memcpy(buffer, labelIds.data(), byteSize);
+
+    output->data = reinterpret_cast<uint8_t*>(buffer);
+    output->dataBytes = byteSize;
+    output->dimsCount = 3;
+    output->dims = (uint64_t*)malloc(output->dimsCount * sizeof(uint64_t));
+    NODE_ASSERT(output->dims != nullptr, "malloc has failed");
+    output->dims[0] = outputBatch;
+    output->dims[1] = 1;
+    output->dims[2] = 1;
+    output->precision = I32;
+    return true;
+}
+
 int initializeInternalManager(void** customNodeLibraryInternalManager, const struct CustomNodeParam* params, int paramsCount) {
     // creating InternalManager instance
     std::unique_ptr<CustomNodeLibraryInternalManager> internalManager = std::make_unique<CustomNodeLibraryInternalManager>();
@@ -348,6 +369,7 @@ int execute(const struct CustomNodeTensor* inputs, int inputsCount, struct Custo
     std::vector<cv::Rect> boxes;
     std::vector<cv::Vec4f> detections;
     std::vector<float> confidences;
+    std::vector<int> labelIds;
 
     for (uint64_t i = 0; i < detectionsCount; i++) {
         float* detection = (float*)(detectionTensor->data + (i * featuresCount * sizeof(float)));
@@ -369,6 +391,7 @@ int execute(const struct CustomNodeTensor* inputs, int inputsCount, struct Custo
             boxes.emplace_back(box);
             detections.emplace_back(detection[3], detection[4], detection[5], detection[6]);
             confidences.emplace_back(confidence);
+            labelIds.emplace_back(labelId);
             if (debugMode) {
                 std::cout << "Detection:\nImageID: " << imageId << "; LabelID:" << labelId << "; Confidence:" << confidence << "; Box:" << box << std::endl;
             }
@@ -413,6 +436,16 @@ int execute(const struct CustomNodeTensor* inputs, int inputsCount, struct Custo
         return 1;
     }
 
+    CustomNodeTensor& labelIdsTensor = (*outputs)[3];
+    labelIdsTensor.name = OUTPUT_LABEL_IDS_TENSOR_NAME;
+    if (!copy_label_ids_into_output(&labelIdsTensor, labelIds)) {
+        cleanup(imagesTensor);
+        cleanup(coordinatesTensor);
+        cleanup(labelIdsTensor);
+        free(*outputs);
+        return 1;
+    }
+
     return 0;
 }
 
@@ -525,6 +558,15 @@ int getOutputsInfo(struct CustomNodeTensorInfo** info, int* infoCount, const str
     (*info)[2].dims[1] = 1;
     (*info)[2].dims[2] = 1;
     (*info)[2].precision = FP32;
+
+    (*info)[3].name = OUTPUT_LABEL_IDS_TENSOR_NAME;
+    (*info)[3].dimsCount = 3;
+    (*info)[3].dims = (uint64_t*)malloc((*info)->dimsCount * sizeof(uint64_t));
+    NODE_ASSERT(((*info)[2].dims) != nullptr, "malloc has failed");
+    (*info)[3].dims[0] = 0;
+    (*info)[3].dims[1] = 1;
+    (*info)[3].dims[2] = 1;
+    (*info)[3].precision = I32;
     return 0;
 }