ImgDetection text label (#1314)

* Added name to ImgDetection

* RVC4 FW: Add name to ImgDetection

* RVC2 FW: Add name to ImgDetection

* RVC2 FW: Fix build

* Address PR comments

* Bump rvc4 fw

Author: asahtik

bindings/python/src/pipeline/datatype/ImgDetectionsBindings.cpp

@@ -36,6 +36,7 @@ void bind_imgdetections(pybind11::module& m, void* pCallstack) {
     // Metadata / raw
     imgDetection.def(py::init<>())
         .def_readwrite("label", &ImgDetection::label)
+        .def_readwrite("name", &ImgDetection::name)
         .def_readwrite("confidence", &ImgDetection::confidence)
         .def_readwrite("xmin", &ImgDetection::xmin)
         .def_readwrite("ymin", &ImgDetection::ymin)

cmake/Depthai/DepthaiDeviceRVC4Config.cmake

@@ -4,4 +4,4 @@ set(DEPTHAI_DEVICE_RVC4_MATURITY "snapshot")
 
 # "version if applicable"
 # set(DEPTHAI_DEVICE_RVC4_VERSION "0.0.1+93f7b75a885aa32f44c5e9f53b74470c49d2b1af")
-set(DEPTHAI_DEVICE_RVC4_VERSION "0.0.1+e0e0596fead6a181e5dea4628058199c79ca892a")
+set(DEPTHAI_DEVICE_RVC4_VERSION "0.0.1+c3853e7a42792202633735a204a1b45758371781")

cmake/Depthai/DepthaiDeviceSideConfig.cmake

@@ -2,7 +2,7 @@
 set(DEPTHAI_DEVICE_SIDE_MATURITY "snapshot")
 
 # "full commit hash of device side binary"
-set(DEPTHAI_DEVICE_SIDE_COMMIT "4f0b0f3219b701b8e5408eea19648a8d0ec06fa7")
+set(DEPTHAI_DEVICE_SIDE_COMMIT "93eeaff81d82368bbe2c326465b1528f8ae93444")
 
 # "version if applicable"
 set(DEPTHAI_DEVICE_SIDE_VERSION "")

include/depthai/common/DetectionParserOptions.hpp

@@ -2,6 +2,7 @@
 
 #include "depthai/common/DetectionNetworkType.hpp"
 #include "depthai/utility/Serialization.hpp"
+#include "depthai/common/optional.hpp"
 
 namespace dai {
 
@@ -18,6 +19,7 @@ struct DetectionParserOptions {
 
     /// YOLO specific network properties
     int classes;
+    std::optional<std::vector<std::string>> classNames;
     int coordinates;
     std::vector<float> anchors;
     std::map<std::string, std::vector<int>> anchorMasks;
@@ -26,6 +28,6 @@ struct DetectionParserOptions {
     float iouThreshold;
 };
 
-DEPTHAI_SERIALIZE_EXT(DetectionParserOptions, nnFamily, subtype, confidenceThreshold, classes, coordinates, anchors, anchorMasks, anchorsV2, iouThreshold);
+DEPTHAI_SERIALIZE_EXT(DetectionParserOptions, nnFamily, subtype, confidenceThreshold, classes, classNames, coordinates, anchors, anchorMasks, anchorsV2, iouThreshold);
 
 }  // namespace dai

include/depthai/pipeline/datatype/ImgDetections.hpp

@@ -11,14 +11,15 @@ namespace dai {
 
 struct ImgDetection {
     uint32_t label = 0;
+    std::string name;
     float confidence = 0.f;
     float xmin = 0.f;
     float ymin = 0.f;
     float xmax = 0.f;
     float ymax = 0.f;
 };
 
-DEPTHAI_SERIALIZE_EXT(ImgDetection, label, confidence, xmin, ymin, xmax, ymax);
+DEPTHAI_SERIALIZE_EXT(ImgDetection, label, name, confidence, xmin, ymin, xmax, ymax);
 
 /**
  * ImgDetections message. Carries normalized detection results

include/depthai/pipeline/node/DetectionParser.hpp

@@ -163,7 +163,6 @@ class DetectionParser : public DeviceNodeCRTP<DeviceNode, DetectionParser, Detec
     void setNNArchiveOther(const NNArchive& nnArchive);
     void setConfig(const dai::NNArchiveVersionedConfig& config);
 
-    std::optional<std::vector<std::string>> mClasses;
     std::optional<NNArchive> mArchive;
 
     std::optional<NNArchiveVersionedConfig> archiveConfig;

include/depthai/properties/DetectionParserProperties.hpp

@@ -4,7 +4,6 @@
 
 #include "depthai/common/DetectionParserOptions.hpp"
 #include "depthai/common/TensorInfo.hpp"
-#include "depthai/common/optional.hpp"
 #include "depthai/properties/Properties.hpp"
 
 namespace dai {

protos/ImgDetections.proto

@@ -11,12 +11,12 @@ message ImgDetections {
     common.ImgTransformation transformation = 5;
 }
 
-
 message ImgDetection {
     uint32 label = 1;
     float confidence = 2;
     float xmin = 3;
     float ymin = 4;
     float xmax = 5;
     float ymax = 6;
+    string name = 7;
 }

src/pipeline/node/DetectionParser.cpp

@@ -212,11 +212,11 @@ int DetectionParser::getNumClasses() const {
 }
 
 std::optional<std::vector<std::string>> DetectionParser::getClasses() const {
-    return mClasses;
+    return properties.parser.classNames;
 }
 
 void DetectionParser::setClasses(const std::vector<std::string>& classes) {
-    mClasses = classes;
+    properties.parser.classNames = classes;
 }
 
 /// Get coordianate size

src/utility/ProtoSerialize.cpp

@@ -260,6 +260,7 @@ std::unique_ptr<google::protobuf::Message> getProtoMessage(const SpatialImgDetec
         // populate SpatialImgDetection.ImgDetection from struct inheritance
         proto::img_detections::ImgDetection* imgDetection = spatialImgDetection->mutable_detection();
         imgDetection->set_label(detection.label);
+        imgDetection->set_name(detection.name);
         imgDetection->set_confidence(detection.confidence);
         imgDetection->set_xmin(detection.xmin);
         imgDetection->set_ymin(detection.ymin);
@@ -381,6 +382,7 @@ std::unique_ptr<google::protobuf::Message> getProtoMessage(const ImgDetections*
     for(const auto& detection : message->detections) {
         proto::img_detections::ImgDetection* imgDetection = imgDetections->add_detections();
         imgDetection->set_label(detection.label);
+        imgDetection->set_name(detection.name);
         imgDetection->set_confidence(detection.confidence);
         imgDetection->set_xmin(detection.xmin);
         imgDetection->set_ymin(detection.ymin);