TICL/plugins/TracksterInferenceByDNN.cc

0001 #include "PhysicsTools/ONNXRuntime/interface/ONNXRuntime.h"
0002 #include "RecoHGCal/TICL/interface/TracksterInferenceByDNN.h"
0003 #include "RecoHGCal/TICL/interface/TracksterInferenceAlgoFactory.h"
0004 #include "FWCore/ParameterSet/interface/ParameterSet.h"
0005 #include "FWCore/Framework/interface/MakerMacros.h"
0006 #include "RecoHGCal/TICL/interface/PatternRecognitionAlgoBase.h"
0007 #include "RecoLocalCalo/HGCalRecAlgos/interface/RecHitTools.h"
0008 #include "TrackstersPCA.h"
0009
0010 namespace ticl {
0011   using namespace cms::Ort;  // Use ONNXRuntime namespace
0012
0013   // Constructor for TracksterInferenceByDNN
0014   TracksterInferenceByDNN::TracksterInferenceByDNN(const edm::ParameterSet& conf)
0015       : TracksterInferenceAlgoBase(conf),
0016         onnxPIDRuntimeInstance_(std::make_unique<cms::Ort::ONNXRuntime>(
0017             conf.getParameter<edm::FileInPath>("onnxPIDModelPath").fullPath().c_str())),
0018         onnxEnergyRuntimeInstance_(std::make_unique<cms::Ort::ONNXRuntime>(
0019             conf.getParameter<edm::FileInPath>("onnxEnergyModelPath").fullPath().c_str())),
0020         inputNames_(conf.getParameter<std::vector<std::string>>("inputNames")),  // Define input names for inference
0021         output_en_(conf.getParameter<std::vector<std::string>>("output_en")),    // Define output energy for inference
0022         output_id_(conf.getParameter<std::vector<std::string>>("output_id")),    // Define output PID for inference
0023         eidMinClusterEnergy_(conf.getParameter<double>("eid_min_cluster_energy")),  // Minimum cluster energy
0024         eidNLayers_(conf.getParameter<int>("eid_n_layers")),                        // Number of layers
0025         eidNClusters_(conf.getParameter<int>("eid_n_clusters")),                    // Number of clusters
0026         doPID_(conf.getParameter<int>("doPID")),                                    // Number of clusters
0027         doRegression_(conf.getParameter<int>("doRegression"))                       // Number of clusters
0028   {
0029     // Initialize ONNX Runtime sessions for PID and Energy models
0030     onnxPIDSession_ = onnxPIDRuntimeInstance_.get();
0031     onnxEnergySession_ = onnxEnergyRuntimeInstance_.get();
0032   }
0033
0034   // Method to process input data and prepare it for inference
0035   void TracksterInferenceByDNN::inputData(const std::vector<reco::CaloCluster>& layerClusters,
0036                                           std::vector<Trackster>& tracksters) {
0037     tracksterIndices_.clear();  // Clear previous indices
0038     for (int i = 0; i < static_cast<int>(tracksters.size()); i++) {
0039       float sumClusterEnergy = 0.;
0040       for (const unsigned int& vertex : tracksters[i].vertices()) {
0041         if (rhtools_.isBarrel(layerClusters[vertex].seed()))
0042           continue;
0043         sumClusterEnergy += static_cast<float>(layerClusters[vertex].energy());
0044         if (sumClusterEnergy >= eidMinClusterEnergy_) {
0045           tracksters[i].setRegressedEnergy(0.f);  // Set regressed energy to 0
0046           tracksters[i].zeroProbabilities();      // Zero out probabilities
0047           tracksterIndices_.push_back(i);         // Add index to the list
0048           break;
0049         }
0050       }
0051     }
0052
0053     // Prepare input shapes and data for inference
0054     batchSize_ = static_cast<int>(tracksterIndices_.size());
0055     if (batchSize_ == 0)
0056       return;  // Exit if no tracksters
0057
0058     std::vector<int64_t> inputShape = {batchSize_, eidNLayers_, eidNClusters_, eidNFeatures_};
0059     input_shapes_ = {inputShape};
0060
0061     input_Data_.clear();
0062     input_Data_.emplace_back(batchSize_ * eidNLayers_ * eidNClusters_ * eidNFeatures_, 0);
0063
0064     for (int i = 0; i < batchSize_; i++) {
0065       const Trackster& trackster = tracksters[tracksterIndices_[i]];
0066
0067       // Prepare indices and sort clusters based on energy
0068       std::vector<int> clusterIndices(trackster.vertices().size());
0069       for (int k = 0; k < static_cast<int>(trackster.vertices().size()); k++) {
0070         clusterIndices[k] = k;
0071       }
0072
0073       std::sort(clusterIndices.begin(), clusterIndices.end(), [&layerClusters, &trackster](const int& a, const int& b) {
0074         return layerClusters[trackster.vertices(a)].energy() > layerClusters[trackster.vertices(b)].energy();
0075       });
0076
0077       std::vector<int> seenClusters(eidNLayers_, 0);
0078
0079       // Fill input data with cluster information
0080       for (const int& k : clusterIndices) {
0081         const reco::CaloCluster& cluster = layerClusters[trackster.vertices(k)];
0082         int j = rhtools_.getLayerWithOffset(cluster.hitsAndFractions()[0].first) - 1;
0083         if (j < eidNLayers_ && seenClusters[j] < eidNClusters_) {
0084           auto index = (i * eidNLayers_ + j) * eidNFeatures_ * eidNClusters_ + seenClusters[j] * eidNFeatures_;
0085           input_Data_[0][index] =
0086               static_cast<float>(cluster.energy() / static_cast<float>(trackster.vertex_multiplicity(k)));
0087           input_Data_[0][index + 1] = static_cast<float>(std::abs(cluster.eta()));
0088           input_Data_[0][index + 2] = static_cast<float>(cluster.phi());
0089           seenClusters[j]++;
0090         }
0091       }
0092     }
0093   }
0094
0095   // Method to run inference and update tracksters
0096   void TracksterInferenceByDNN::runInference(std::vector<Trackster>& tracksters) {
0097     if (batchSize_ == 0)
0098       return;  // Exit if no batch
0099
0100     if (doPID_ and doRegression_) {
0101       // Run energy model inference
0102       auto result = onnxEnergySession_->run(inputNames_, input_Data_, input_shapes_, output_en_, batchSize_);
0103       auto& energyOutputTensor = result[0];
0104       if (!output_en_.empty()) {
0105         for (int i = 0; i < static_cast<int>(batchSize_); i++) {
0106           const float energy = energyOutputTensor[i];
0107           tracksters[tracksterIndices_[i]].setRegressedEnergy(energy);  // Update energy
0108         }
0109       }
0110     }
0111
0112     if (doPID_) {
0113       // Run PID model inference
0114       auto pidOutput = onnxPIDSession_->run(inputNames_, input_Data_, input_shapes_, output_id_, batchSize_);
0115       auto pidOutputTensor = pidOutput[0];
0116       float* probs = pidOutputTensor.data();
0117       if (!output_id_.empty()) {
0118         for (int i = 0; i < batchSize_; i++) {
0119           tracksters[tracksterIndices_[i]].setProbabilities(probs);             // Update probabilities
0120           probs += tracksters[tracksterIndices_[i]].id_probabilities().size();  // Move to next set of probabilities
0121         }
0122       }
0123     }
0124   }
0125   // Method to fill parameter set description for configuration
0126   void TracksterInferenceByDNN::fillPSetDescription(edm::ParameterSetDescription& iDesc) {
0127     iDesc.add<int>("algo_verbosity", 0);
0128     iDesc
0129         .add<edm::FileInPath>(
0130             "onnxPIDModelPath",
0131             edm::FileInPath("RecoHGCal/TICL/data/ticlv5/onnx_models/DNN/patternrecognition/id_v0.onnx"))
0132         ->setComment("Path to ONNX PID model CLU3D");
0133     iDesc
0134         .add<edm::FileInPath>(
0135             "onnxEnergyModelPath",
0136             edm::FileInPath("RecoHGCal/TICL/data/ticlv5/onnx_models/DNN/patternrecognition/energy_v0.onnx"))
0137         ->setComment("Path to ONNX Energy model CLU3D");
0138     iDesc.add<std::vector<std::string>>("inputNames", {"input"});
0139     iDesc.add<std::vector<std::string>>("output_en", {"enreg_output"});
0140     iDesc.add<std::vector<std::string>>("output_id", {"pid_output"});
0141     iDesc.add<double>("eid_min_cluster_energy", 1.0);
0142     iDesc.add<int>("eid_n_layers", 50);
0143     iDesc.add<int>("eid_n_clusters", 10);
0144     iDesc.add<int>("doPID", 1);
0145     iDesc.add<int>("doRegression", 1);
0146   }
0147 }  // namespace ticl