opencv-ruby 1.0

data/ext/opencv2/dnn/all_layers-rb.hpp

@@ -0,0 +1,4 @@
+#include <rice/rice.hpp>
+#include <rice/stl.hpp>
+
+void Init_AllLayers();

data/ext/opencv2/dnn/dict-rb.cpp

@@ -0,0 +1,57 @@
+#include <opencv2/dnn/dict.hpp>
+#include "dict-rb.hpp"
+
+using namespace Rice;
+
+Rice::Class rb_cCvDnnDict;
+Rice::Class rb_cCvDnnDictValue;
+
+void Init_Dict()
+{
+  Module rb_mCv = define_module("Cv");
+  Module rb_mCvDnn = define_module_under(rb_mCv, "Dnn");
+  
+  rb_cCvDnnDictValue = define_class_under<cv::dnn::DictValue>(rb_mCvDnn, "DictValue").
+    define_constructor(Constructor<cv::dnn::DictValue, const cv::dnn::DictValue&>(),
+      Arg("r")).
+    define_constructor(Constructor<cv::dnn::DictValue, bool>(),
+      Arg("i")).
+    define_constructor(Constructor<cv::dnn::DictValue, int64>(),
+      Arg("i") = static_cast<int64>(0)).
+    define_constructor(Constructor<cv::dnn::DictValue, int>(),
+      Arg("i")).
+    define_constructor(Constructor<cv::dnn::DictValue, unsigned int>(),
+      Arg("p")).
+    define_constructor(Constructor<cv::dnn::DictValue, double>(),
+      Arg("p")).
+    define_constructor(Constructor<cv::dnn::DictValue, const cv::String&>(),
+      Arg("s")).
+    define_constructor(Constructor<cv::dnn::DictValue, const char*>(),
+      Arg("s")).
+    define_method("size", &cv::dnn::DictValue::size).
+    define_method("int?", &cv::dnn::DictValue::isInt).
+    define_method("string?", &cv::dnn::DictValue::isString).
+    define_method("real?", &cv::dnn::DictValue::isReal).
+    define_method("get_int_value", &cv::dnn::DictValue::getIntValue,
+      Arg("idx") = static_cast<int>(-1)).
+    define_method("get_real_value", &cv::dnn::DictValue::getRealValue,
+      Arg("idx") = static_cast<int>(-1)).
+    define_method("get_string_value", &cv::dnn::DictValue::getStringValue,
+      Arg("idx") = static_cast<int>(-1)).
+    define_method("assign", &cv::dnn::DictValue::operator=,
+      Arg("r"));
+  
+  rb_cCvDnnDict = define_class_under<cv::dnn::Dict>(rb_mCvDnn, "Dict").
+    define_constructor(Constructor<cv::dnn::Dict>()).
+    define_method("has?", &cv::dnn::Dict::has,
+      Arg("key")).
+    define_method<cv::dnn::DictValue*(cv::dnn::Dict::*)(const cv::String&)>("ptr", &cv::dnn::Dict::ptr,
+      Arg("key")).
+    define_method<const cv::dnn::DictValue*(cv::dnn::Dict::*)(const cv::String&) const>("ptr", &cv::dnn::Dict::ptr,
+      Arg("key")).
+  //  define_method("get", &cv::dnn::Dict::get,
+   //   Arg("key")).
+    define_method("erase", &cv::dnn::Dict::erase,
+      Arg("key"));
+  //  define_iterator<std::map::const_iterator(cv::dnn::Dict::*)() const>(&cv::dnn::Dict::begin, &cv::dnn::Dict::end, "each_const");
+}

data/ext/opencv2/dnn/dict-rb.hpp

@@ -0,0 +1,4 @@
+#include <rice/rice.hpp>
+#include <rice/stl.hpp>
+
+void Init_Dict();

data/ext/opencv2/dnn/dnn-rb.cpp

@@ -0,0 +1,548 @@
+#include <opencv2/dnn/dnn.hpp>
+#include "../core/cvstd_wrapper-rb.hpp"
+#include "dnn-rb.hpp"
+
+using namespace Rice;
+
+//Rice::Class rb_cCvDnnBackendWrapper;
+Rice::Class rb_cCvDnnClassificationModel;
+Rice::Class rb_cCvDnnDetectionModel;
+Rice::Class rb_cCvDnnImage2BlobParams;
+Rice::Class rb_cCvDnnKeypointsModel;
+Rice::Class rb_cCvDnnLayer;
+Rice::Class rb_cCvDnnLayerParams;
+Rice::Class rb_cCvDnnModel;
+Rice::Class rb_cCvDnnNet;
+Rice::Class rb_cCvDnnSegmentationModel;
+Rice::Class rb_cCvDnnTextDetectionModel;
+Rice::Class rb_cCvDnnTextDetectionModelDB;
+Rice::Class rb_cCvDnnTextDetectionModelEAST;
+Rice::Class rb_cCvDnnTextRecognitionModel;
+Rice::Class rb_cMatShape;
+
+void Init_Dnn()
+{
+  Module rb_mCv = define_module("Cv");
+  Module rb_mCvDnn = define_module_under(rb_mCv, "Dnn");
+  
+  Enum<cv::dnn::Backend> rb_cCvDnnBackend = define_enum_under<cv::dnn::Backend>("Backend", rb_mCvDnn).
+    define_value("DNN_BACKEND_DEFAULT", cv::dnn::Backend::DNN_BACKEND_DEFAULT).
+    define_value("DNN_BACKEND_HALIDE", cv::dnn::Backend::DNN_BACKEND_HALIDE).
+    define_value("DNN_BACKEND_INFERENCE_ENGINE", cv::dnn::Backend::DNN_BACKEND_INFERENCE_ENGINE).
+    define_value("DNN_BACKEND_OPENCV", cv::dnn::Backend::DNN_BACKEND_OPENCV).
+    define_value("DNN_BACKEND_VKCOM", cv::dnn::Backend::DNN_BACKEND_VKCOM).
+    define_value("DNN_BACKEND_CUDA", cv::dnn::Backend::DNN_BACKEND_CUDA).
+    define_value("DNN_BACKEND_WEBNN", cv::dnn::Backend::DNN_BACKEND_WEBNN).
+    define_value("DNN_BACKEND_TIMVX", cv::dnn::Backend::DNN_BACKEND_TIMVX).
+    define_value("DNN_BACKEND_CANN", cv::dnn::Backend::DNN_BACKEND_CANN);
+  
+  Enum<cv::dnn::Target> rb_cCvDnnTarget = define_enum_under<cv::dnn::Target>("Target", rb_mCvDnn).
+    define_value("DNN_TARGET_CPU", cv::dnn::Target::DNN_TARGET_CPU).
+    define_value("DNN_TARGET_OPENCL", cv::dnn::Target::DNN_TARGET_OPENCL).
+    define_value("DNN_TARGET_OPENCL_FP16", cv::dnn::Target::DNN_TARGET_OPENCL_FP16).
+    define_value("DNN_TARGET_MYRIAD", cv::dnn::Target::DNN_TARGET_MYRIAD).
+    define_value("DNN_TARGET_VULKAN", cv::dnn::Target::DNN_TARGET_VULKAN).
+    define_value("DNN_TARGET_FPGA", cv::dnn::Target::DNN_TARGET_FPGA).
+    define_value("DNN_TARGET_CUDA", cv::dnn::Target::DNN_TARGET_CUDA).
+    define_value("DNN_TARGET_CUDA_FP16", cv::dnn::Target::DNN_TARGET_CUDA_FP16).
+    define_value("DNN_TARGET_HDDL", cv::dnn::Target::DNN_TARGET_HDDL).
+    define_value("DNN_TARGET_NPU", cv::dnn::Target::DNN_TARGET_NPU).
+    define_value("DNN_TARGET_CPU_FP16", cv::dnn::Target::DNN_TARGET_CPU_FP16);
+  
+  Enum<cv::dnn::DataLayout> rb_cCvDnnDataLayout = define_enum_under<cv::dnn::DataLayout>("DataLayout", rb_mCvDnn).
+    define_value("DNN_LAYOUT_UNKNOWN", cv::dnn::DataLayout::DNN_LAYOUT_UNKNOWN).
+    define_value("DNN_LAYOUT_ND", cv::dnn::DataLayout::DNN_LAYOUT_ND).
+    define_value("DNN_LAYOUT_NCHW", cv::dnn::DataLayout::DNN_LAYOUT_NCHW).
+    define_value("DNN_LAYOUT_NCDHW", cv::dnn::DataLayout::DNN_LAYOUT_NCDHW).
+    define_value("DNN_LAYOUT_NHWC", cv::dnn::DataLayout::DNN_LAYOUT_NHWC).
+    define_value("DNN_LAYOUT_NDHWC", cv::dnn::DataLayout::DNN_LAYOUT_NDHWC).
+    define_value("DNN_LAYOUT_PLANAR", cv::dnn::DataLayout::DNN_LAYOUT_PLANAR);
+  
+  rb_mCvDnn.define_module_function("get_available_backends", &cv::dnn::getAvailableBackends);
+  
+  rb_mCvDnn.define_module_function("get_available_targets", &cv::dnn::getAvailableTargets,
+    Arg("be"));
+  
+  rb_mCvDnn.define_module_function("enable_model_diagnostics", &cv::dnn::enableModelDiagnostics,
+    Arg("is_diagnostics_mode"));
+  
+  rb_cCvDnnLayerParams = define_class_under<cv::dnn::LayerParams, cv::dnn::Dict>(rb_mCvDnn, "LayerParams").
+    define_constructor(Constructor<cv::dnn::LayerParams>()).
+    define_attr("blobs", &cv::dnn::LayerParams::blobs).
+    define_attr("name", &cv::dnn::LayerParams::name).
+    define_attr("type", &cv::dnn::LayerParams::type);
+  
+  //usr/share/ruby/bundled_gems.rb:82:in 'Kernel.require': /usr/local/src/opencv-ruby/lib/opencv_ruby.so: undefined symbol: _ZTIN2cv3dnn14dnn4_v2024122314BackendWrapperE - /usr/local/src/opencv-ruby/lib/opencv_ruby.so (LoadError)
+  /*rb_cCvDnnBackendWrapper = define_class_under<cv::dnn::BackendWrapper>(rb_mCvDnn, "BackendWrapper").
+    define_method("copy_to_host", &cv::dnn::BackendWrapper::copyToHost).
+    define_method("set_host_dirty", &cv::dnn::BackendWrapper::setHostDirty).
+    define_attr("backend_id", &cv::dnn::BackendWrapper::backendId).
+    define_attr("target_id", &cv::dnn::BackendWrapper::targetId);
+  */
+  rb_cCvDnnLayer = define_class_under<cv::dnn::Layer, cv::Algorithm>(rb_mCvDnn, "Layer").
+    define_attr("blobs", &cv::dnn::Layer::blobs).
+    define_method<void(cv::dnn::Layer::*)(cv::InputArrayOfArrays, cv::OutputArrayOfArrays)>("finalize", &cv::dnn::Layer::finalize,
+      Arg("inputs"), Arg("outputs")).
+    define_method<void(cv::dnn::Layer::*)(cv::InputArrayOfArrays, cv::OutputArrayOfArrays, cv::OutputArrayOfArrays)>("forward", &cv::dnn::Layer::forward,
+      Arg("inputs"), Arg("outputs"), Arg("internals")).
+    define_method("try_quantize", &cv::dnn::Layer::tryQuantize,
+      Arg("scales"), Arg("zeropoints"), Arg("params")).
+    define_method("forward_fallback", &cv::dnn::Layer::forward_fallback,
+      Arg("inputs"), Arg("outputs"), Arg("internals")).
+    define_method("input_name_to_index", &cv::dnn::Layer::inputNameToIndex,
+      Arg("input_name")).
+    define_method("output_name_to_index", &cv::dnn::Layer::outputNameToIndex,
+      Arg("output_name")).
+    define_method("support_backend", &cv::dnn::Layer::supportBackend,
+      Arg("backend_id")).
+
+  /* These all return BackendNode which is not a public api 
+    define_method("init_halide", &cv::dnn::Layer::initHalide,
+      Arg("inputs")).
+    define_method("init_ngraph", &cv::dnn::Layer::initNgraph,
+      Arg("inputs"), Arg("nodes")).
+    define_method("init_vk_com", &cv::dnn::Layer::initVkCom,
+      Arg("inputs"), Arg("outputs")).
+    define_method("init_webnn", &cv::dnn::Layer::initWebnn,
+      Arg("inputs"), Arg("nodes")).
+    define_method("init_cuda", &cv::dnn::Layer::initCUDA,
+      Arg("context"), Arg("inputs"), Arg("outputs")).
+    define_method("init_tim_vx", &cv::dnn::Layer::initTimVX,
+      Arg("tim_vx_info"), Arg("inputs_wrapper"), Arg("outputs_wrapper"), Arg("is_last")).
+    define_method("init_cann", &cv::dnn::Layer::initCann,
+      Arg("inputs"), Arg("outputs"), Arg("nodes")).
+    define_method("apply_halide_scheduler", &cv::dnn::Layer::applyHalideScheduler,
+      Arg("node"), Arg("inputs"), Arg("outputs"), Arg("target_id")).
+    define_method("try_attach", &cv::dnn::Layer::tryAttach,
+      Arg("node")).*/
+//    define_method("set_activation", &cv::dnn::Layer::setActivation,
+//      Arg("layer")).
+    define_method("try_fuse", &cv::dnn::Layer::tryFuse,
+      Arg("top")).
+    define_method("get_scale_shift", &cv::dnn::Layer::getScaleShift,
+      Arg("scale"), Arg("shift")).
+    define_method("get_scale_zeropoint", &cv::dnn::Layer::getScaleZeropoint,
+      Arg("scale"), Arg("zeropoint")).
+    define_method("unset_attached", &cv::dnn::Layer::unsetAttached).
+    define_method("get_memory_shapes", &cv::dnn::Layer::getMemoryShapes,
+      Arg("inputs"), Arg("required_outputs"), Arg("outputs"), Arg("internals")).
+    define_method("get_flops", &cv::dnn::Layer::getFLOPS,
+      Arg("inputs"), Arg("outputs")).
+    define_method("update_memory_shapes", &cv::dnn::Layer::updateMemoryShapes,
+      Arg("inputs")).
+    define_attr("name", &cv::dnn::Layer::name).
+    define_attr("type", &cv::dnn::Layer::type).
+    define_attr("preferable_target", &cv::dnn::Layer::preferableTarget).
+    define_constructor(Constructor<cv::dnn::Layer>()).
+    define_constructor(Constructor<cv::dnn::Layer, const cv::dnn::LayerParams&>(),
+      Arg("params")).
+    define_method("set_params_from", &cv::dnn::Layer::setParamsFrom,
+      Arg("params"));
+  
+  rb_cCvDnnNet = define_class_under<cv::dnn::Net>(rb_mCvDnn, "Net").
+    define_constructor(Constructor<cv::dnn::Net>()).
+    define_singleton_function<cv::dnn::Net(*)(const cv::String&, const cv::String&)>("read_from_model_optimizer", &cv::dnn::Net::readFromModelOptimizer,
+      Arg("xml"), Arg("bin")).
+    define_singleton_function<cv::dnn::Net(*)(const std::vector<unsigned char>&, const std::vector<unsigned char>&)>("read_from_model_optimizer", &cv::dnn::Net::readFromModelOptimizer,
+      Arg("buffer_model_config"), Arg("buffer_weights")).
+    define_singleton_function<cv::dnn::Net(*)(const uchar*, size_t, const uchar*, size_t)>("read_from_model_optimizer", &cv::dnn::Net::readFromModelOptimizer,
+      Arg("buffer_model_config_ptr"), Arg("buffer_model_config_size"), Arg("buffer_weights_ptr"), Arg("buffer_weights_size")).
+    define_method("empty?", &cv::dnn::Net::empty).
+    define_method("dump", &cv::dnn::Net::dump).
+    define_method("dump_to_file", &cv::dnn::Net::dumpToFile,
+      Arg("path")).
+    define_method("dump_to_pbtxt", &cv::dnn::Net::dumpToPbtxt,
+      Arg("path")).
+    define_method<int(cv::dnn::Net::*)(const cv::String&, const cv::String&, const int&, cv::dnn::LayerParams&)>("add_layer", &cv::dnn::Net::addLayer,
+      Arg("name"), Arg("type"), Arg("dtype"), Arg("params")).
+    define_method<int(cv::dnn::Net::*)(const cv::String&, const cv::String&, cv::dnn::LayerParams&)>("add_layer", &cv::dnn::Net::addLayer,
+      Arg("name"), Arg("type"), Arg("params")).
+    define_method<int(cv::dnn::Net::*)(const cv::String&, const cv::String&, const int&, cv::dnn::LayerParams&)>("add_layer_to_prev", &cv::dnn::Net::addLayerToPrev,
+      Arg("name"), Arg("type"), Arg("dtype"), Arg("params")).
+    define_method<int(cv::dnn::Net::*)(const cv::String&, const cv::String&, cv::dnn::LayerParams&)>("add_layer_to_prev", &cv::dnn::Net::addLayerToPrev,
+      Arg("name"), Arg("type"), Arg("params")).
+    define_method("get_layer_id", &cv::dnn::Net::getLayerId,
+      Arg("layer")).
+    define_method("get_layer_names", &cv::dnn::Net::getLayerNames).
+    define_method<cv::Ptr<cv::dnn::Layer>(cv::dnn::Net::*)(int) const>("get_layer", &cv::dnn::Net::getLayer,
+      Arg("layer_id")).
+    define_method<cv::Ptr<cv::dnn::Layer>(cv::dnn::Net::*)(const cv::String&) const>("get_layer", &cv::dnn::Net::getLayer,
+      Arg("layer_name")).
+    define_method<cv::Ptr<cv::dnn::Layer>(cv::dnn::Net::*)(const cv::dnn::Net::LayerId&) const>("get_layer", &cv::dnn::Net::getLayer,
+      Arg("layer_id")).
+    define_method("get_layer_inputs", &cv::dnn::Net::getLayerInputs,
+      Arg("layer_id")).
+    define_method<void(cv::dnn::Net::*)(cv::String, cv::String)>("connect", &cv::dnn::Net::connect,
+      Arg("out_pin"), Arg("inp_pin")).
+    define_method<void(cv::dnn::Net::*)(int, int, int, int)>("connect", &cv::dnn::Net::connect,
+      Arg("out_layer_id"), Arg("out_num"), Arg("inp_layer_id"), Arg("inp_num")).
+    define_method("register_output", &cv::dnn::Net::registerOutput,
+      Arg("output_name"), Arg("layer_id"), Arg("output_port")).
+    define_method("set_inputs_names", &cv::dnn::Net::setInputsNames,
+      Arg("input_blob_names")).
+    define_method("set_input_shape", &cv::dnn::Net::setInputShape,
+      Arg("input_name"), Arg("shape")).
+    define_method<cv::Mat(cv::dnn::Net::*)(const cv::String&)>("forward", &cv::dnn::Net::forward,
+      Arg("output_name") = static_cast<const String &>(String())).
+    define_method("forward_async", &cv::dnn::Net::forwardAsync,
+      Arg("output_name") = static_cast<const String &>(String())).
+    define_method<void(cv::dnn::Net::*)(cv::OutputArrayOfArrays, const cv::String&)>("forward", &cv::dnn::Net::forward,
+      Arg("output_blobs"), Arg("output_name") = static_cast<const String &>(String())).
+    define_method<void(cv::dnn::Net::*)(cv::OutputArrayOfArrays, const std::vector<std::string>&)>("forward", &cv::dnn::Net::forward,
+      Arg("output_blobs"), Arg("out_blob_names")).
+    define_method<void(cv::dnn::Net::*)(std::vector<std::vector<cv::Mat>>&, const std::vector<std::string>&)>("forward", &cv::dnn::Net::forward,
+      Arg("output_blobs"), Arg("out_blob_names")).
+    define_method("quantize", &cv::dnn::Net::quantize,
+      Arg("calib_data"), Arg("inputs_dtype"), Arg("outputs_dtype"), Arg("per_channel") = static_cast<bool>(true)).
+    define_method("get_input_details", &cv::dnn::Net::getInputDetails,
+      Arg("scales"), Arg("zeropoints")).
+    define_method("get_output_details", &cv::dnn::Net::getOutputDetails,
+      Arg("scales"), Arg("zeropoints")).
+    define_method("set_halide_scheduler", &cv::dnn::Net::setHalideScheduler,
+      Arg("scheduler")).
+    define_method("set_preferable_backend", &cv::dnn::Net::setPreferableBackend,
+      Arg("backend_id")).
+    define_method("set_preferable_target", &cv::dnn::Net::setPreferableTarget,
+      Arg("target_id")).
+    define_method("set_input", &cv::dnn::Net::setInput,
+      Arg("blob"), Arg("name") = static_cast<const cv::String &>(""), Arg("scalefactor") = static_cast<double>(1.0), Arg("mean") = static_cast<const cv::Scalar &>(cv::Scalar())).
+    define_method<void(cv::dnn::Net::*)(int, int, const cv::Mat&)>("set_param", &cv::dnn::Net::setParam,
+      Arg("layer"), Arg("num_param"), Arg("blob")).
+    define_method<void(cv::dnn::Net::*)(const cv::String&, int, const cv::Mat&)>("set_param", &cv::dnn::Net::setParam,
+      Arg("layer_name"), Arg("num_param"), Arg("blob")).
+    define_method<cv::Mat(cv::dnn::Net::*)(int, int) const>("get_param", &cv::dnn::Net::getParam,
+      Arg("layer"), Arg("num_param") = static_cast<int>(0)).
+    define_method<cv::Mat(cv::dnn::Net::*)(const cv::String&, int) const>("get_param", &cv::dnn::Net::getParam,
+      Arg("layer_name"), Arg("num_param") = static_cast<int>(0)).
+    define_method("get_unconnected_out_layers", &cv::dnn::Net::getUnconnectedOutLayers).
+    define_method("get_unconnected_out_layers_names", &cv::dnn::Net::getUnconnectedOutLayersNames).
+    define_method<void(cv::dnn::Net::*)(const std::vector<std::vector<int>>&, std::vector<int>&, std::vector<std::vector<std::vector<int>>>&, std::vector<std::vector<std::vector<int>>>&) const>("get_layers_shapes", &cv::dnn::Net::getLayersShapes,
+      Arg("net_input_shapes"), Arg("layers_ids"), Arg("in_layers_shapes"), Arg("out_layers_shapes")).
+    define_method<void(cv::dnn::Net::*)(const cv::dnn::MatShape&, std::vector<int>&, std::vector<std::vector<std::vector<int>>>&, std::vector<std::vector<std::vector<int>>>&) const>("get_layers_shapes", &cv::dnn::Net::getLayersShapes,
+      Arg("net_input_shape"), Arg("layers_ids"), Arg("in_layers_shapes"), Arg("out_layers_shapes")).
+    define_method<void(cv::dnn::Net::*)(const cv::dnn::MatShape&, const int, std::vector<std::vector<int>>&, std::vector<std::vector<int>>&) const>("get_layer_shapes", &cv::dnn::Net::getLayerShapes,
+      Arg("net_input_shape"), Arg("layer_id"), Arg("in_layer_shapes"), Arg("out_layer_shapes")).
+    define_method<void(cv::dnn::Net::*)(const std::vector<std::vector<int>>&, const int, std::vector<std::vector<int>>&, std::vector<std::vector<int>>&) const>("get_layer_shapes", &cv::dnn::Net::getLayerShapes,
+      Arg("net_input_shapes"), Arg("layer_id"), Arg("in_layer_shapes"), Arg("out_layer_shapes")).
+    define_method<int64(cv::dnn::Net::*)(const std::vector<std::vector<int>>&) const>("get_flops", &cv::dnn::Net::getFLOPS,
+      Arg("net_input_shapes")).
+    define_method<int64(cv::dnn::Net::*)(const cv::dnn::MatShape&) const>("get_flops", &cv::dnn::Net::getFLOPS,
+      Arg("net_input_shape")).
+    define_method<int64(cv::dnn::Net::*)(const int, const std::vector<std::vector<int>>&) const>("get_flops", &cv::dnn::Net::getFLOPS,
+      Arg("layer_id"), Arg("net_input_shapes")).
+    define_method<int64(cv::dnn::Net::*)(const int, const cv::dnn::MatShape&) const>("get_flops", &cv::dnn::Net::getFLOPS,
+      Arg("layer_id"), Arg("net_input_shape")).
+    define_method("get_layer_types", &cv::dnn::Net::getLayerTypes,
+      Arg("layers_types")).
+    define_method("get_layers_count", &cv::dnn::Net::getLayersCount,
+      Arg("layer_type")).
+    define_method<void(cv::dnn::Net::*)(const std::vector<std::vector<int>>&, size_t&, size_t&) const>("get_memory_consumption", &cv::dnn::Net::getMemoryConsumption,
+      Arg("net_input_shapes"), Arg("weights"), Arg("blobs")).
+    define_method<void(cv::dnn::Net::*)(const cv::dnn::MatShape&, size_t&, size_t&) const>("get_memory_consumption", &cv::dnn::Net::getMemoryConsumption,
+      Arg("net_input_shape"), Arg("weights"), Arg("blobs")).
+    define_method<void(cv::dnn::Net::*)(const int, const std::vector<std::vector<int>>&, size_t&, size_t&) const>("get_memory_consumption", &cv::dnn::Net::getMemoryConsumption,
+      Arg("layer_id"), Arg("net_input_shapes"), Arg("weights"), Arg("blobs")).
+    define_method<void(cv::dnn::Net::*)(const int, const cv::dnn::MatShape&, size_t&, size_t&) const>("get_memory_consumption", &cv::dnn::Net::getMemoryConsumption,
+      Arg("layer_id"), Arg("net_input_shape"), Arg("weights"), Arg("blobs")).
+    // Doesn't compile on Fedora with OpenCV 4.11.0 but code hasn't changed in 2 years. Confused.
+    //define_method<void(cv::dnn::Net::*)(const std::vector<std::vector<int>>&, std::vector<int>&, std::vector<unsigned long long>&, std::vector<unsigned long long>&) const>("get_memory_consumption", &cv::dnn::Net::getMemoryConsumption,
+    //  Arg("net_input_shapes"), Arg("layer_ids"), Arg("weights"), Arg("blobs")).
+    //define_method<void(cv::dnn::Net::*)(const cv::dnn::MatShape&, std::vector<int>&, std::vector<unsigned long long>&, std::vector<unsigned long long>&) const>("get_memory_consumption", &cv::dnn::Net::getMemoryConsumption,
+    // Arg("net_input_shape"), Arg("layer_ids"), Arg("weights"), Arg("blobs")).
+    define_method("enable_fusion", &cv::dnn::Net::enableFusion,
+      Arg("fusion")).
+    define_method("enable_winograd", &cv::dnn::Net::enableWinograd,
+      Arg("use_winograd")).
+    define_method("get_perf_profile", &cv::dnn::Net::getPerfProfile,
+      Arg("timings"));
+    //define_method("get_impl", &cv::dnn::Net::getImpl).
+    //define_method("get_impl_ref", &cv::dnn::Net::getImplRef);
+  
+  rb_mCvDnn.define_module_function<cv::dnn::Net(*)(const cv::String&, const cv::String&)>("read_net_from_darknet", &cv::dnn::readNetFromDarknet,
+    Arg("cfg_file"), Arg("darknet_model") = static_cast<const String &>(String()));
+  
+  rb_mCvDnn.define_module_function<cv::dnn::Net(*)(const std::vector<unsigned char>&, const std::vector<unsigned char>&)>("read_net_from_darknet", &cv::dnn::readNetFromDarknet,
+    Arg("buffer_cfg"), Arg("buffer_model") = static_cast<const std::vector<uchar> &>(std::vector<uchar>()));
+  
+  rb_mCvDnn.define_module_function<cv::dnn::Net(*)(const char*, size_t, const char*, size_t)>("read_net_from_darknet", &cv::dnn::readNetFromDarknet,
+    Arg("buffer_cfg"), Arg("len_cfg"), Arg("buffer_model") = static_cast<const char *>(NULL), Arg("len_model") = static_cast<size_t>(0));
+  
+  rb_mCvDnn.define_module_function<cv::dnn::Net(*)(const cv::String&, const cv::String&)>("read_net_from_caffe", &cv::dnn::readNetFromCaffe,
+    Arg("prototxt"), Arg("caffe_model") = static_cast<const String &>(String()));
+  
+  rb_mCvDnn.define_module_function<cv::dnn::Net(*)(const std::vector<unsigned char>&, const std::vector<unsigned char>&)>("read_net_from_caffe", &cv::dnn::readNetFromCaffe,
+    Arg("buffer_proto"), Arg("buffer_model") = static_cast<const std::vector<uchar> &>(std::vector<uchar>()));
+  
+  rb_mCvDnn.define_module_function<cv::dnn::Net(*)(const char*, size_t, const char*, size_t)>("read_net_from_caffe", &cv::dnn::readNetFromCaffe,
+    Arg("buffer_proto"), Arg("len_proto"), Arg("buffer_model") = static_cast<const char *>(NULL), Arg("len_model") = static_cast<size_t>(0));
+  
+  rb_mCvDnn.define_module_function<cv::dnn::Net(*)(const cv::String&, const cv::String&)>("read_net_from_tensorflow", &cv::dnn::readNetFromTensorflow,
+    Arg("model"), Arg("config") = static_cast<const String &>(String()));
+  
+  rb_mCvDnn.define_module_function<cv::dnn::Net(*)(const std::vector<unsigned char>&, const std::vector<unsigned char>&)>("read_net_from_tensorflow", &cv::dnn::readNetFromTensorflow,
+    Arg("buffer_model"), Arg("buffer_config") = static_cast<const std::vector<uchar> &>(std::vector<uchar>()));
+  
+  rb_mCvDnn.define_module_function<cv::dnn::Net(*)(const char*, size_t, const char*, size_t)>("read_net_from_tensorflow", &cv::dnn::readNetFromTensorflow,
+    Arg("buffer_model"), Arg("len_model"), Arg("buffer_config") = static_cast<const char *>(NULL), Arg("len_config") = static_cast<size_t>(0));
+  
+  rb_mCvDnn.define_module_function<cv::dnn::Net(*)(const cv::String&)>("read_net_from_tf_lite", &cv::dnn::readNetFromTFLite,
+    Arg("model"));
+  
+  rb_mCvDnn.define_module_function<cv::dnn::Net(*)(const std::vector<unsigned char>&)>("read_net_from_tf_lite", &cv::dnn::readNetFromTFLite,
+    Arg("buffer_model"));
+  
+  rb_mCvDnn.define_module_function<cv::dnn::Net(*)(const char*, size_t)>("read_net_from_tf_lite", &cv::dnn::readNetFromTFLite,
+    Arg("buffer_model"), Arg("len_model"));
+  
+  rb_mCvDnn.define_module_function("read_net_from_torch", &cv::dnn::readNetFromTorch,
+    Arg("model"), Arg("is_binary") = static_cast<bool>(true), Arg("evaluate") = static_cast<bool>(true));
+  
+  rb_mCvDnn.define_module_function<cv::dnn::Net(*)(const cv::String&, const cv::String&, const cv::String&)>("read_net", &cv::dnn::readNet,
+    Arg("model"), Arg("config") = static_cast<const String &>(""), Arg("framework") = static_cast<const String &>(""));
+  
+  rb_mCvDnn.define_module_function<cv::dnn::Net(*)(const cv::String&, const std::vector<unsigned char>&, const std::vector<unsigned char>&)>("read_net", &cv::dnn::readNet,
+    Arg("framework"), Arg("buffer_model"), Arg("buffer_config") = static_cast<const std::vector<uchar> &>(std::vector<uchar>()));
+  
+  rb_mCvDnn.define_module_function("read_torch_blob", &cv::dnn::readTorchBlob,
+    Arg("filename"), Arg("is_binary") = static_cast<bool>(true));
+  
+  rb_mCvDnn.define_module_function<cv::dnn::Net(*)(const cv::String&, const cv::String&)>("read_net_from_model_optimizer", &cv::dnn::readNetFromModelOptimizer,
+    Arg("xml"), Arg("bin") = static_cast<const String &>(""));
+  
+  rb_mCvDnn.define_module_function<cv::dnn::Net(*)(const std::vector<unsigned char>&, const std::vector<unsigned char>&)>("read_net_from_model_optimizer", &cv::dnn::readNetFromModelOptimizer,
+    Arg("buffer_model_config"), Arg("buffer_weights"));
+  
+  rb_mCvDnn.define_module_function<cv::dnn::Net(*)(const uchar*, size_t, const uchar*, size_t)>("read_net_from_model_optimizer", &cv::dnn::readNetFromModelOptimizer,
+    Arg("buffer_model_config_ptr"), Arg("buffer_model_config_size"), Arg("buffer_weights_ptr"), Arg("buffer_weights_size"));
+  
+  rb_mCvDnn.define_module_function<cv::dnn::Net(*)(const cv::String&)>("read_net_from_onnx", &cv::dnn::readNetFromONNX,
+    Arg("onnx_file"));
+  
+  rb_mCvDnn.define_module_function<cv::dnn::Net(*)(const char*, size_t)>("read_net_from_onnx", &cv::dnn::readNetFromONNX,
+    Arg("buffer"), Arg("size_buffer"));
+  
+  rb_mCvDnn.define_module_function<cv::dnn::Net(*)(const std::vector<unsigned char>&)>("read_net_from_onnx", &cv::dnn::readNetFromONNX,
+    Arg("buffer"));
+  
+  rb_mCvDnn.define_module_function("read_tensor_from_onnx", &cv::dnn::readTensorFromONNX,
+    Arg("path"));
+  
+  rb_mCvDnn.define_module_function<cv::Mat(*)(cv::InputArray, double, const cv::Size&, const cv::Scalar&, bool, bool, int)>("blob_from_image", &cv::dnn::blobFromImage,
+    Arg("image"), Arg("scalefactor") = static_cast<double>(1.0), Arg("size") = static_cast<const cv::Size &> (cv::Size()), Arg("mean") = static_cast<const cv::Scalar &>(cv::Scalar()), Arg("swap_rb") = static_cast<bool>(false), Arg("crop") = static_cast<bool>(false), Arg("ddepth") = static_cast<int>(CV_32F));
+  
+  rb_mCvDnn.define_module_function<void(*)(cv::InputArray, cv::OutputArray, double, const cv::Size&, const cv::Scalar&, bool, bool, int)>("blob_from_image", &cv::dnn::blobFromImage,
+    Arg("image"), Arg("blob"), Arg("scalefactor") = static_cast<double>(1.0), Arg("size") = static_cast<const cv::Size &> (cv::Size()), Arg("mean") = static_cast<const cv::Scalar &>(cv::Scalar()), Arg("swap_rb") = static_cast<bool>(false), Arg("crop") = static_cast<bool>(false), Arg("ddepth") = static_cast<int>(CV_32F));
+  
+  rb_mCvDnn.define_module_function<cv::Mat(*)(cv::InputArrayOfArrays, double, cv::Size, const cv::Scalar&, bool, bool, int)>("blob_from_images", &cv::dnn::blobFromImages,
+    Arg("images"), Arg("scalefactor") = static_cast<double>(1.0), Arg("size") = static_cast<cv::Size> (cv::Size()), Arg("mean") = static_cast<const cv::Scalar &>(cv::Scalar()), Arg("swap_rb") = static_cast<bool>(false), Arg("crop") = static_cast<bool>(false), Arg("ddepth") = static_cast<int>(CV_32F));
+  
+  rb_mCvDnn.define_module_function<void(*)(cv::InputArrayOfArrays, cv::OutputArray, double, cv::Size, const cv::Scalar&, bool, bool, int)>("blob_from_images", &cv::dnn::blobFromImages,
+    Arg("images"), Arg("blob"), Arg("scalefactor") = static_cast<double>(1.0), Arg("size") = static_cast<cv::Size> (cv::Size()), Arg("mean") = static_cast<const cv::Scalar &>(cv::Scalar()), Arg("swap_rb") = static_cast<bool>(false), Arg("crop") = static_cast<bool>(false), Arg("ddepth") = static_cast<int>(CV_32F));
+  
+  Enum<cv::dnn::ImagePaddingMode> rb_cCvDnnImagePaddingMode = define_enum_under<cv::dnn::ImagePaddingMode>("ImagePaddingMode", rb_mCvDnn).
+    define_value("DNN_PMODE_NULL", cv::dnn::ImagePaddingMode::DNN_PMODE_NULL).
+    define_value("DNN_PMODE_CROP_CENTER", cv::dnn::ImagePaddingMode::DNN_PMODE_CROP_CENTER).
+    define_value("DNN_PMODE_LETTERBOX", cv::dnn::ImagePaddingMode::DNN_PMODE_LETTERBOX);
+  
+  rb_cCvDnnImage2BlobParams = define_class_under<cv::dnn::Image2BlobParams>(rb_mCvDnn, "Image2BlobParams").
+    define_constructor(Constructor<cv::dnn::Image2BlobParams>()).
+    define_constructor(Constructor<cv::dnn::Image2BlobParams, const cv::Scalar&, const cv::Size&, const cv::Scalar&, bool, int, cv::dnn::DataLayout, cv::dnn::ImagePaddingMode, cv::Scalar>(),
+      Arg("scalefactor"), Arg("size") = static_cast<const cv::Size &> (cv::Size()), Arg("mean") = static_cast<const cv::Scalar &>(cv::Scalar()), Arg("swap_rb") = static_cast<bool>(false), Arg("ddepth") = static_cast<int>(CV_32F), Arg("datalayout") = static_cast<cv::dnn::DataLayout>(cv::dnn::DNN_LAYOUT_NCHW), Arg("mode") = static_cast<cv::dnn::ImagePaddingMode>(cv::dnn::DNN_PMODE_NULL), Arg("border_value") = static_cast<cv::Scalar>(0.0)).
+    define_attr("scalefactor", &cv::dnn::Image2BlobParams::scalefactor).
+    define_attr("size", &cv::dnn::Image2BlobParams::size).
+    define_attr("mean", &cv::dnn::Image2BlobParams::mean).
+    define_attr("swap_rb", &cv::dnn::Image2BlobParams::swapRB).
+    define_attr("ddepth", &cv::dnn::Image2BlobParams::ddepth).
+    define_attr("datalayout", &cv::dnn::Image2BlobParams::datalayout).
+    define_attr("paddingmode", &cv::dnn::Image2BlobParams::paddingmode).
+    define_attr("border_value", &cv::dnn::Image2BlobParams::borderValue).
+    define_method("blob_rect_to_image_rect", &cv::dnn::Image2BlobParams::blobRectToImageRect,
+      Arg("r_blob"), Arg("size")).
+    define_method("blob_rects_to_image_rects", &cv::dnn::Image2BlobParams::blobRectsToImageRects,
+      Arg("r_blob"), Arg("r_img"), Arg("size"));
+  
+  rb_mCvDnn.define_module_function<cv::Mat(*)(cv::InputArray, const cv::dnn::Image2BlobParams&)>("blob_from_image_with_params", &cv::dnn::blobFromImageWithParams,
+    Arg("image"), Arg("param") = static_cast<const cv::dnn::Image2BlobParams &>(cv::dnn::Image2BlobParams()));
+  
+  rb_mCvDnn.define_module_function<void(*)(cv::InputArray, cv::OutputArray, const cv::dnn::Image2BlobParams&)>("blob_from_image_with_params", &cv::dnn::blobFromImageWithParams,
+    Arg("image"), Arg("blob"), Arg("param") = static_cast<const cv::dnn::Image2BlobParams &>(cv::dnn::Image2BlobParams()));
+  
+  rb_mCvDnn.define_module_function<cv::Mat(*)(cv::InputArrayOfArrays, const cv::dnn::Image2BlobParams&)>("blob_from_images_with_params", &cv::dnn::blobFromImagesWithParams,
+    Arg("images"), Arg("param") = static_cast<const cv::dnn::Image2BlobParams &>(cv::dnn::Image2BlobParams()));
+  
+  rb_mCvDnn.define_module_function<void(*)(cv::InputArrayOfArrays, cv::OutputArray, const cv::dnn::Image2BlobParams&)>("blob_from_images_with_params", &cv::dnn::blobFromImagesWithParams,
+    Arg("images"), Arg("blob"), Arg("param") = static_cast<const cv::dnn::Image2BlobParams &>(cv::dnn::Image2BlobParams()));
+  
+  rb_mCvDnn.define_module_function("images_from_blob", &cv::dnn::imagesFromBlob,
+    Arg("blob_"), Arg("images_"));
+  
+  rb_mCvDnn.define_module_function("shrink_caffe_model", &cv::dnn::shrinkCaffeModel,
+    Arg("src"), Arg("dst"), Arg("layers_types") = static_cast<const std::vector<String> &>(std::vector<String>()));
+  
+  rb_mCvDnn.define_module_function("write_text_graph", &cv::dnn::writeTextGraph,
+    Arg("model"), Arg("output"));
+  
+  rb_mCvDnn.define_module_function<void(*)(const std::vector<cv::Rect_<int>>&, const std::vector<float>&, const float, const float, std::vector<int>&, const float, const int)>("nms_boxes", &cv::dnn::NMSBoxes,
+    Arg("bboxes"), Arg("scores"), Arg("score_threshold"), Arg("nms_threshold"), Arg("indices"), Arg("eta") = static_cast<const float>(1.f), Arg("top_k") = static_cast<const int>(0));
+  
+  rb_mCvDnn.define_module_function<void(*)(const std::vector<cv::Rect_<double>>&, const std::vector<float>&, const float, const float, std::vector<int>&, const float, const int)>("nms_boxes", &cv::dnn::NMSBoxes,
+    Arg("bboxes"), Arg("scores"), Arg("score_threshold"), Arg("nms_threshold"), Arg("indices"), Arg("eta") = static_cast<const float>(1.f), Arg("top_k") = static_cast<const int>(0));
+  
+  rb_mCvDnn.define_module_function<void(*)(const std::vector<cv::RotatedRect>&, const std::vector<float>&, const float, const float, std::vector<int>&, const float, const int)>("nms_boxes", &cv::dnn::NMSBoxes,
+    Arg("bboxes"), Arg("scores"), Arg("score_threshold"), Arg("nms_threshold"), Arg("indices"), Arg("eta") = static_cast<const float>(1.f), Arg("top_k") = static_cast<const int>(0));
+  
+  rb_mCvDnn.define_module_function<void(*)(const std::vector<cv::Rect_<int>>&, const std::vector<float>&, const std::vector<int>&, const float, const float, std::vector<int>&, const float, const int)>("nms_boxes_batched", &cv::dnn::NMSBoxesBatched,
+    Arg("bboxes"), Arg("scores"), Arg("class_ids"), Arg("score_threshold"), Arg("nms_threshold"), Arg("indices"), Arg("eta") = static_cast<const float>(1.f), Arg("top_k") = static_cast<const int>(0));
+  
+  rb_mCvDnn.define_module_function<void(*)(const std::vector<cv::Rect_<double>>&, const std::vector<float>&, const std::vector<int>&, const float, const float, std::vector<int>&, const float, const int)>("nms_boxes_batched", &cv::dnn::NMSBoxesBatched,
+    Arg("bboxes"), Arg("scores"), Arg("class_ids"), Arg("score_threshold"), Arg("nms_threshold"), Arg("indices"), Arg("eta") = static_cast<const float>(1.f), Arg("top_k") = static_cast<const int>(0));
+  
+  Enum<cv::dnn::SoftNMSMethod> rb_cCvDnnSoftNMSMethod = define_enum_under<cv::dnn::SoftNMSMethod>("SoftNMSMethod", rb_mCvDnn).
+    define_value("SOFTNMS_LINEAR", cv::dnn::SoftNMSMethod::SOFTNMS_LINEAR).
+    define_value("SOFTNMS_GAUSSIAN", cv::dnn::SoftNMSMethod::SOFTNMS_GAUSSIAN);
+  
+  rb_mCvDnn.define_module_function("soft_nms_boxes", &cv::dnn::softNMSBoxes,
+    Arg("bboxes"), Arg("scores"), Arg("updated_scores"), Arg("score_threshold"), Arg("nms_threshold"), Arg("indices"), Arg("top_k") = static_cast<size_t>(0), Arg("sigma") = static_cast<const float>(0.5), Arg("method") = static_cast<cv::dnn::SoftNMSMethod>(cv::dnn::SoftNMSMethod::SOFTNMS_GAUSSIAN));
+  
+  rb_cCvDnnModel = define_class_under<cv::dnn::Model>(rb_mCvDnn, "Model").
+    define_constructor(Constructor<cv::dnn::Model, const cv::dnn::Model&>(),
+      Arg("")).
+    define_method<cv::dnn::Model&(cv::dnn::Model::*)(const cv::dnn::Model&)>("assign", &cv::dnn::Model::operator=,
+      Arg("")).
+    define_method<cv::dnn::Model&(cv::dnn::Model::*)(cv::dnn::Model&&)>("assign", &cv::dnn::Model::operator=,
+      Arg("")).
+    define_constructor(Constructor<cv::dnn::Model, const cv::String&, const cv::String&>(),
+      Arg("model"), Arg("config") = static_cast<const cv::String &>("")).
+    define_constructor(Constructor<cv::dnn::Model, const cv::dnn::Net&>(),
+      Arg("network")).
+    define_method<cv::dnn::Model&(cv::dnn::Model::*)(const cv::Size&)>("set_input_size", &cv::dnn::Model::setInputSize,
+      Arg("size")).
+    define_method<cv::dnn::Model&(cv::dnn::Model::*)(int, int)>("set_input_size", &cv::dnn::Model::setInputSize,
+      Arg("width"), Arg("height")).
+    define_method("set_input_mean", &cv::dnn::Model::setInputMean,
+      Arg("mean")).
+    define_method("set_input_scale", &cv::dnn::Model::setInputScale,
+      Arg("scale")).
+    define_method("set_input_crop", &cv::dnn::Model::setInputCrop,
+      Arg("crop")).
+    define_method("set_input_swap_rb", &cv::dnn::Model::setInputSwapRB,
+      Arg("swap_rb")).
+    define_method("set_output_names", &cv::dnn::Model::setOutputNames,
+      Arg("out_names")).
+    define_method("set_input_params", &cv::dnn::Model::setInputParams,
+      Arg("scale") = static_cast<double>(1.0), Arg("size") = static_cast<const cv::Size &> (cv::Size()), Arg("mean") = static_cast<const cv::Scalar &>(cv::Scalar()), Arg("swap_rb") = static_cast<bool>(false), Arg("crop") = static_cast<bool>(false)).
+    define_method("predict", &cv::dnn::Model::predict,
+      Arg("frame"), Arg("outs")).
+    define_method("set_preferable_backend", &cv::dnn::Model::setPreferableBackend,
+      Arg("backend_id")).
+    define_method("set_preferable_target", &cv::dnn::Model::setPreferableTarget,
+      Arg("target_id")).
+    define_method("enable_winograd", &cv::dnn::Model::enableWinograd,
+      Arg("use_winograd")).
+//    define_method("to_cv/dnn/dnn4_v20250619/net &", [](const cv::dnn::Model& self) -> cv::dnn::Net&
+ //   {
+  //    return self;
+   // }).
+    define_method<cv::dnn::Net&(cv::dnn::Model::*)() const>("get_network_", &cv::dnn::Model::getNetwork_).
+    define_method<cv::dnn::Net&(cv::dnn::Model::*)()>("get_network_", &cv::dnn::Model::getNetwork_);
+    //define_method("get_impl", &cv::dnn::Model::getImpl).
+    //define_method("get_impl_ref", &cv::dnn::Model::getImplRef);
+
+  rb_cCvDnnClassificationModel = define_class_under<cv::dnn::ClassificationModel, cv::dnn::Model>(rb_mCvDnn, "ClassificationModel").
+    define_constructor(Constructor<cv::dnn::ClassificationModel, const cv::String&, const cv::String&>(),
+      Arg("model"), Arg("config") = static_cast<const cv::String &>("")).
+    define_constructor(Constructor<cv::dnn::ClassificationModel, const cv::dnn::Net&>(),
+      Arg("network")).
+    define_method("set_enable_softmax_post_processing", &cv::dnn::ClassificationModel::setEnableSoftmaxPostProcessing,
+      Arg("enable")).
+    define_method("get_enable_softmax_post_processing", &cv::dnn::ClassificationModel::getEnableSoftmaxPostProcessing).
+    define_method<std::pair<int, float>(cv::dnn::ClassificationModel::*)(cv::InputArray)>("classify", &cv::dnn::ClassificationModel::classify,
+      Arg("frame")).
+    define_method<void(cv::dnn::ClassificationModel::*)(cv::InputArray, int&, float&)>("classify", &cv::dnn::ClassificationModel::classify,
+      Arg("frame"), Arg("class_id"), Arg("conf"));
+  
+  rb_cCvDnnKeypointsModel = define_class_under<cv::dnn::KeypointsModel, cv::dnn::Model>(rb_mCvDnn, "KeypointsModel").
+    define_constructor(Constructor<cv::dnn::KeypointsModel, const cv::String&, const cv::String&>(),
+      Arg("model"), Arg("config") = static_cast<const cv::String &>("")).
+    define_constructor(Constructor<cv::dnn::KeypointsModel, const cv::dnn::Net&>(),
+      Arg("network")).
+    define_method("estimate", &cv::dnn::KeypointsModel::estimate,
+      Arg("frame"), Arg("thresh") = static_cast<float>(0.5));
+  
+  rb_cCvDnnSegmentationModel = define_class_under<cv::dnn::SegmentationModel, cv::dnn::Model>(rb_mCvDnn, "SegmentationModel").
+    define_constructor(Constructor<cv::dnn::SegmentationModel, const cv::String&, const cv::String&>(),
+      Arg("model"), Arg("config") = static_cast<const cv::String &>("")).
+    define_constructor(Constructor<cv::dnn::SegmentationModel, const cv::dnn::Net&>(),
+      Arg("network")).
+    define_method("segment", &cv::dnn::SegmentationModel::segment,
+      Arg("frame"), Arg("mask"));
+  
+  rb_cCvDnnDetectionModel = define_class_under<cv::dnn::DetectionModel, cv::dnn::Model>(rb_mCvDnn, "DetectionModel").
+    define_constructor(Constructor<cv::dnn::DetectionModel, const cv::String&, const cv::String&>(),
+      Arg("model"), Arg("config") = static_cast<const cv::String &>("")).
+    define_constructor(Constructor<cv::dnn::DetectionModel, const cv::dnn::Net&>(),
+      Arg("network")).
+    define_method("set_nms_across_classes", &cv::dnn::DetectionModel::setNmsAcrossClasses,
+      Arg("value")).
+    define_method("get_nms_across_classes", &cv::dnn::DetectionModel::getNmsAcrossClasses).
+    define_method("detect", &cv::dnn::DetectionModel::detect,
+      Arg("frame"), Arg("class_ids"), Arg("confidences"), Arg("boxes"), Arg("conf_threshold") = static_cast<float>(0.5f), Arg("nms_threshold") = static_cast<float>(0.0f));
+  
+  rb_cCvDnnTextRecognitionModel = define_class_under<cv::dnn::TextRecognitionModel, cv::dnn::Model>(rb_mCvDnn, "TextRecognitionModel").
+    define_constructor(Constructor<cv::dnn::TextRecognitionModel, const cv::dnn::Net&>(),
+      Arg("network")).
+    define_constructor(Constructor<cv::dnn::TextRecognitionModel, const std::string&, const std::string&>(),
+      Arg("model"), Arg("config") = static_cast<const std::string &>("")).
+    define_method("set_decode_type", &cv::dnn::TextRecognitionModel::setDecodeType,
+      Arg("decode_type")).
+    define_method("get_decode_type", &cv::dnn::TextRecognitionModel::getDecodeType).
+    define_method("set_decode_opts_ctc_prefix_beam_search", &cv::dnn::TextRecognitionModel::setDecodeOptsCTCPrefixBeamSearch,
+      Arg("beam_size"), Arg("voc_prune_size") = static_cast<int>(0)).
+    define_method("set_vocabulary", &cv::dnn::TextRecognitionModel::setVocabulary,
+      Arg("vocabulary")).
+    define_method("get_vocabulary", &cv::dnn::TextRecognitionModel::getVocabulary).
+    define_method<std::string(cv::dnn::TextRecognitionModel::*)(cv::InputArray) const>("recognize", &cv::dnn::TextRecognitionModel::recognize,
+      Arg("frame")).
+    define_method<void(cv::dnn::TextRecognitionModel::*)(cv::InputArray, cv::InputArrayOfArrays, std::vector<std::string>&) const>("recognize", &cv::dnn::TextRecognitionModel::recognize,
+      Arg("frame"), Arg("roi_rects"), Arg("results"));
+  
+  rb_cCvDnnTextDetectionModel = define_class_under<cv::dnn::TextDetectionModel, cv::dnn::Model>(rb_mCvDnn, "TextDetectionModel").
+    define_method<void(cv::dnn::TextDetectionModel::*)(cv::InputArray, std::vector<std::vector<cv::Point_<int>>>&, std::vector<float>&) const>("detect", &cv::dnn::TextDetectionModel::detect,
+      Arg("frame"), Arg("detections"), Arg("confidences")).
+    define_method<void(cv::dnn::TextDetectionModel::*)(cv::InputArray, std::vector<std::vector<cv::Point_<int>>>&) const>("detect", &cv::dnn::TextDetectionModel::detect,
+      Arg("frame"), Arg("detections")).
+    define_method<void(cv::dnn::TextDetectionModel::*)(cv::InputArray, std::vector<cv::RotatedRect>&, std::vector<float>&) const>("detect_text_rectangles", &cv::dnn::TextDetectionModel::detectTextRectangles,
+      Arg("frame"), Arg("detections"), Arg("confidences")).
+    define_method<void(cv::dnn::TextDetectionModel::*)(cv::InputArray, std::vector<cv::RotatedRect>&) const>("detect_text_rectangles", &cv::dnn::TextDetectionModel::detectTextRectangles,
+      Arg("frame"), Arg("detections"));
+  
+  rb_cCvDnnTextDetectionModelEAST = define_class_under<cv::dnn::TextDetectionModel_EAST, cv::dnn::TextDetectionModel>(rb_mCvDnn, "TextDetectionModelEAST").
+    define_constructor(Constructor<cv::dnn::TextDetectionModel_EAST, const cv::dnn::Net&>(),
+      Arg("network")).
+    define_constructor(Constructor<cv::dnn::TextDetectionModel_EAST, const std::string&, const std::string&>(),
+      Arg("model"), Arg("config") = static_cast<const std::string &>("")).
+    define_method("set_confidence_threshold", &cv::dnn::TextDetectionModel_EAST::setConfidenceThreshold,
+      Arg("conf_threshold")).
+    define_method("get_confidence_threshold", &cv::dnn::TextDetectionModel_EAST::getConfidenceThreshold).
+    define_method("set_nms_threshold", &cv::dnn::TextDetectionModel_EAST::setNMSThreshold,
+      Arg("nms_threshold")).
+    define_method("get_nms_threshold", &cv::dnn::TextDetectionModel_EAST::getNMSThreshold);
+  
+  rb_cCvDnnTextDetectionModelDB = define_class_under<cv::dnn::TextDetectionModel_DB, cv::dnn::TextDetectionModel>(rb_mCvDnn, "TextDetectionModelDB").
+    define_constructor(Constructor<cv::dnn::TextDetectionModel_DB, const cv::dnn::Net&>(),
+      Arg("network")).
+    define_constructor(Constructor<cv::dnn::TextDetectionModel_DB, const std::string&, const std::string&>(),
+      Arg("model"), Arg("config") = static_cast<const std::string &>("")).
+    define_method("set_binary_threshold", &cv::dnn::TextDetectionModel_DB::setBinaryThreshold,
+      Arg("binary_threshold")).
+    define_method("get_binary_threshold", &cv::dnn::TextDetectionModel_DB::getBinaryThreshold).
+    define_method("set_polygon_threshold", &cv::dnn::TextDetectionModel_DB::setPolygonThreshold,
+      Arg("polygon_threshold")).
+    define_method("get_polygon_threshold", &cv::dnn::TextDetectionModel_DB::getPolygonThreshold).
+    define_method("set_unclip_ratio", &cv::dnn::TextDetectionModel_DB::setUnclipRatio,
+      Arg("unclip_ratio")).
+    define_method("get_unclip_ratio", &cv::dnn::TextDetectionModel_DB::getUnclipRatio).
+    define_method("set_max_candidates", &cv::dnn::TextDetectionModel_DB::setMaxCandidates,
+      Arg("max_candidates")).
+    define_method("get_max_candidates", &cv::dnn::TextDetectionModel_DB::getMaxCandidates);
+}