Created new visualizer, fixed sonata tags: #219

examples/mnist-client/CMakeLists.txt

@@ -1,32 +1,24 @@
-#[[
-© 2024 AO Kaspersky Lab
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-]]
-
+#[[
+© 2024 AO Kaspersky Lab
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+]]
+
 cmake_minimum_required(VERSION 3.25)
 
 set(CMAKE_CXX_STANDARD 17)
 set(CMAKE_CXX_EXTENSIONS OFF)
 
-find_package(OpenCV)
-if (NOT OpenCV_FOUND)
-    message(WARNING "OpenCV was not found. Skipping MNIST example...")
-    return()
-endif()
-
-include_directories(${OpenCV_INCLUDE_DIRS})
-
 if(CMAKE_VERSION VERSION_GREATER_EQUAL "3.30")
     # Suppress Boost warning.
     cmake_policy(SET CMP0167 OLD)
@@ -45,4 +37,9 @@ add_executable(mnist-client
     process_data.cpp
 )
 
-target_link_libraries(mnist-client PRIVATE KNP::BaseFramework::Core ${OpenCV_LIBS} ${Boost_LIBRARIES})
+find_package(PkgConfig REQUIRED)
+pkg_check_modules(GRAPHVIZ REQUIRED IMPORTED_TARGET libgvc libcgraph)
+
+target_include_directories(mnist-client PRIVATE ${GRAPHVIZ_INCLUDE_DIRS})
+target_link_libraries(mnist-client PRIVATE PkgConfig::GRAPHVIZ  )
+target_link_libraries(mnist-client PRIVATE KNP::BaseFramework::Core ${Boost_LIBRARIES} )

examples/mnist-client/main.cpp

@@ -74,15 +74,9 @@ int main(int argc, char **argv)
         // https://click.kaspersky.com/?hl=en-US&version=2.0&pid=KNP&link=online_help&helpid=235801
         // https://click.kaspersky.com/?hl=en-US&version=2.0&pid=KNP&link=online_help&helpid=274991
         knp::framework::Network network = knp::framework::sonata::load_network(network_path);
-        // Constructs a network graph.
-        const knp::framework::NetworkGraph net_graph(network);
-        // Prints descriptions of graph connections.
-        knp::framework::print_network_description(net_graph);
-        // Draws a subgraph in the OpenCV window.
-        // Press `Esc` to exit the OpenCV window.
-        knp::framework::position_network_test(
-            knp::framework::NetworkGraph(network), knp::framework::divide_graph_by_connectivity(net_graph)[0],
-            {1000, 700});
+
+        // knp::framework::set_saving_path("temp_test_dir");
+        knp::framework::visualize_network(network);
     }
 
     // If `task=infer`, the function loads a network and runs inference.

examples/mnist-client/process_data.cpp

@@ -27,38 +27,13 @@
 #include <utility>
 #include <vector>
 
-#include <opencv2/core.hpp>
-#include <opencv2/highgui.hpp>
-
 
 /**
  * @brief Filesystem namespace alias.
  */
 namespace fs = std::filesystem;
 
 
-/**
- * @brief Visualize data instances.
- * @param data vector of flattened images.
- * @param size output image size.
- */
-cv::Mat draw_data(const std::vector<unsigned char> &data, const cv::Size &size)
-{
-    cv::Mat out_img(size, CV_8UC1);
-    for (int y = 0; y < size.height; ++y)
-    {
-        for (int x = 0; x < size.width; ++x)
-        {
-            out_img.at<unsigned char>(cv::Point2i{x, y}) = data[size.width * y + x];
-        }
-    }
-    cv::imshow("Data", out_img);
-    std::cout << cv::sum(out_img) << std::endl;
-    cv::waitKey(0);
-    return out_img;
-}
-
-
 /**
  * @brief Turn data frames into spikes.
  * @param buf data frame.

examples/mnist-learn/CMakeLists.txt

@@ -26,6 +26,8 @@ if (NOT TARGET KNP::BaseFramework::Core)
 endif()
 
 find_package(Boost ${KNP_BOOST_MIN_VERSION} COMPONENTS program_options REQUIRED)
+find_package(PkgConfig REQUIRED)
+pkg_check_modules(GRAPHVIZ REQUIRED IMPORTED_TARGET libgvc libcgraph)
 
 add_executable(${PROJECT_NAME}
     main.cpp
@@ -40,4 +42,7 @@ add_executable(${PROJECT_NAME}
     models/altai/prepare_network_for_inference.cpp
     models/altai/spike_generators.cpp)
 target_include_directories(${PROJECT_NAME} PRIVATE ${PROJECT_SOURCE_DIR})
-target_link_libraries(${PROJECT_NAME} PRIVATE KNP::BaseFramework::Core Boost::program_options)
+target_link_libraries(${PROJECT_NAME} PRIVATE
+    KNP::BaseFramework::Core
+    Boost::program_options
+    )

examples/mnist-learn/main.cpp

@@ -19,8 +19,16 @@
  * limitations under the License.
  */
 
+#include <knp/framework/io/input_channel.h>
+#include <knp/framework/monitoring/observer.h>
+#include <knp/framework/network.h>
+#include <knp/framework/sonata/network_io.h>
+#include <knp/framework/visualizer/visualize_network.h>
+
 #include <iostream>
 
+#include <boost/program_options.hpp>
+
 #include "dataset.h"
 #include "evaluate_results.h"
 #include "inference.h"
@@ -31,6 +39,7 @@
 
 /**
  * @brief Run whole model.
+ *
  * @tparam Neuron Neuron type.
  * @param model_desc Model description.
  */
@@ -43,10 +52,16 @@ void run_model(const ModelDescription& model_desc)
 
     // Online Help link: https://click.kaspersky.com/?hl=en-US&version=2.0&pid=KNP&link=online_help&helpid=243548
     knp::framework::BackendLoader backend_loader;
-    train_model<Neuron>(model_desc, dataset, network, backend_loader);
+    auto backend = train_model<Neuron>(model_desc, dataset, network, backend_loader);
 
     if (!model_desc.model_saving_path_.empty()) save_network(model_desc, network);
 
+
+    // knp::framework::set_saving_path("temp_test_dir");
+    visualize_network(network.network_);
+    visualize_network(network.network_, backend);
+    visualize_bus(network.network_, backend);
+
     auto inference_spikes = infer_model<Neuron>(model_desc, dataset, network, backend_loader);
 
     evaluate_results(inference_spikes, dataset);
@@ -55,8 +70,10 @@ void run_model(const ModelDescription& model_desc)
 
 /**
  * @brief Main function.
+ *
  * @param argc Argument count.
  * @param argv Arguments value.
+ *
  * @return Error code.
  */
 int main(int argc, char** argv)

examples/mnist-learn/models/blifat/construct_network.cpp

@@ -48,19 +48,19 @@ using ResourceNeuronData = knp::neuron_traits::neuron_parameters<knp::neuron_tra
 struct NetworkPopulations
 {
     /// Input population.
-    const PopulationInfo &input_pop_;
+    const PopulationInfo& input_pop_;
     /// Output population.
-    const PopulationInfo &output_pop_;
+    const PopulationInfo& output_pop_;
     /// Gate population. Used for training.
-    const PopulationInfo &gate_pop_;
+    const PopulationInfo& gate_pop_;
     /// Population for rasterized images.
-    const PopulationInfo &raster_pop_;
+    const PopulationInfo& raster_pop_;
     /// Population for images labels.
-    const PopulationInfo &target_pop_;
+    const PopulationInfo& target_pop_;
 };
 
 
-static NetworkPopulations create_populations(NetworkConstructor &constructor)
+static NetworkPopulations create_populations(NetworkConstructor& constructor)
 {
     // Creating neurons.
     // Online Help link: https://click.kaspersky.com/?hl=en-US&version=2.0&pid=KNP&link=online_help&helpid=235859
@@ -78,25 +78,22 @@ static NetworkPopulations create_populations(NetworkConstructor &constructor)
     input_neuron.stochastic_stimulation_ = stochastic_stimulation;
 
     // Creating populations using neurons.
-    const auto &input_pop =
+    const auto& input_pop =
         constructor.add_population(input_neuron, num_input_neurons, PopulationRole::INPUT, true, "INPUT");
-    const auto &output_pop =
+    const auto& output_pop =
         constructor.add_population(default_neuron, classes_amount, PopulationRole::OUTPUT, true, "OUTPUT");
-    const auto &gate_pop =
-        constructor.add_population(default_neuron, classes_amount, PopulationRole::NORMAL, false, "GATE");
-    const auto &raster_pop = constructor.add_channeled_population(input_size, true);
-    const auto &target_pop = constructor.add_channeled_population(classes_amount, false);
-
+    const auto& gate_pop =
+        constructor.add_population(default_neuron, classes_amount, PopulationRole::NORMAL, true, "GATE");  // was false
+    const auto& raster_pop = constructor.add_channeled_population(input_size, true);
+    const auto& target_pop = constructor.add_channeled_population(classes_amount, true);  // was false
     // Returning them.
     return {input_pop, output_pop, gate_pop, raster_pop, target_pop};
 }
 
 
 static void create_projections(
-    AnnotatedNetwork &network, NetworkConstructor &constructor, const NetworkPopulations &pops)
+    AnnotatedNetwork& network, NetworkConstructor& constructor, const NetworkPopulations& pops)
 {
-    // Creating synapse and projection out of it. Multiple times.
-
     // Synapse creation.
     ResourceSynapseParams raster_to_input_synapse;
     raster_to_input_synapse.rule_.synaptic_resource_ =
@@ -172,7 +169,7 @@ static void create_projections(
  * @see [Online Help](https://click.kaspersky.com/?hl=en-US&version=2.0&pid=KNP&link=online_help&helpid=235801)
  */
 template <>
-AnnotatedNetwork construct_network<knp::neuron_traits::BLIFATNeuron>(const ModelDescription &model_desc)
+AnnotatedNetwork construct_network<knp::neuron_traits::BLIFATNeuron>(const ModelDescription& model_desc)
 {
     AnnotatedNetwork result;
 
@@ -187,7 +184,6 @@ AnnotatedNetwork construct_network<knp::neuron_traits::BLIFATNeuron>(const Model
 
         // Add input_pop as WTA sender.
         result.data_.wta_data_.emplace_back().first.push_back(pops.input_pop_.uid_);
-
         create_projections(result, constructor, pops);
     }
 

examples/mnist-learn/training.h

@@ -26,10 +26,10 @@
 #include <knp/framework/projection/wta.h>
 #include <knp/framework/tags/name.h>
 
-#include <map>
-#include <memory>
 #include <filesystem>
 #include <fstream>
+#include <map>
+#include <memory>
 #include <string>
 #include <utility>
 #include <vector>
@@ -162,12 +162,13 @@ void train_network(
  * @param backend_loader Backend loader.
  */
 template <typename Neuron>
-void train_model(
+std::shared_ptr<knp::core::Backend> train_model(
     const ModelDescription& model_desc, const Dataset& dataset, AnnotatedNetwork& network,
     knp::framework::BackendLoader& backend_loader)
 {
     std::shared_ptr<knp::core::Backend> training_backend = backend_loader.load(model_desc.training_backend_path_);
     train_network<Neuron>(training_backend, network, model_desc, dataset);
 
     prepare_network_for_inference<Neuron>(training_backend, model_desc, network);
+    return training_backend;
 }

knp/base-framework/CMakeLists.txt

@@ -34,7 +34,6 @@ project(knp-base-framework VERSION "${KNP_VERSION}" LANGUAGES C CXX
 set(${PROJECT_NAME}_PUBLIC_INCLUDE_DIR "knp/framework")
 
 find_package(Boost ${KNP_BOOST_MIN_VERSION} COMPONENTS filesystem REQUIRED)
-find_package(OpenCV REQUIRED)
 
 include(GNUInstallDirs)
 include(knp-functions)
@@ -77,13 +76,15 @@ knp_add_library("${PROJECT_NAME}-core"
     impl/inference_evaluation/classification/processor.cpp
     impl/observer.cpp
     impl/logging.cpp
-    impl/visualizer/graph_physics.cpp
     impl/visualizer/visualize_network.cpp
+    impl/visualizer/graph_helpers.cpp
+    impl/visualizer/dot_helpers.cpp
+    impl/visualizer/generate_files.cpp
     ${${PROJECT_NAME}_headers}
     ALIAS KNP::BaseFramework::Core
     LINK_PRIVATE
         spdlog::spdlog_header_only Boost::headers Boost::filesystem HighFive hdf5-static csv2
-        ${OpenCV_LIBS}
+        ${GRAPHVIZ_LIBRARIES}
         # Hack to build with CLang.
         ${ADD_LIBS}
     LINK_PUBLIC

knp/base-framework/impl/sonata/load_network.cpp

@@ -99,16 +99,16 @@ std::vector<core::AllPopulationsVariant> load_populations(const fs::path &pop_h5
 
     for (size_t i = 0; i < num_populations; ++i)
     {
-        const std::string proj_name = group.getObjectName(i);
+        const std::string pop_name = group.getObjectName(i);
         const int neuron_type =
-            group.getGroup(proj_name).getDataSet("node_type_id").read<std::vector<int>>()[0];  // One type only.
+            group.getGroup(pop_name).getDataSet("node_type_id").read<std::vector<int>>()[0];  // One type only.
         // Check if type is in type_file.
         if (neuron_type == get_neuron_type_id<neuron_traits::BLIFATNeuron>())
-            result.emplace_back(load_population<neuron_traits::BLIFATNeuron>(group, proj_name));
+            result.emplace_back(load_population<neuron_traits::BLIFATNeuron>(group, pop_name));
         else if (neuron_type == get_neuron_type_id<neuron_traits::SynapticResourceSTDPBLIFATNeuron>())
-            result.emplace_back(load_population<neuron_traits::SynapticResourceSTDPBLIFATNeuron>(group, proj_name));
+            result.emplace_back(load_population<neuron_traits::SynapticResourceSTDPBLIFATNeuron>(group, pop_name));
         else if (neuron_type == get_neuron_type_id<neuron_traits::AltAILIF>())
-            result.emplace_back(load_population<neuron_traits::AltAILIF>(group, proj_name));
+            result.emplace_back(load_population<neuron_traits::AltAILIF>(group, pop_name));
         // TODO: Add other supported types or better use a template.
     }
     return result;