Package org.jlab.rec.alert.AI


package org.jlab.rec.alert.AI
  • Classes
    Class
    Description
     
    Orchestrates GNN-based track finding: builds the graph, runs the exported edge scorer, extracts tracks via connected components on edge scores thresholded at 0.1, and converts each node-set back into a TrackCandidate carrying per-superlayer Clusters so the downstream helix fit / Kalman stages can consume it.
     
    Model of What.
    DJL wrapper around the GravNet TorchScript model exported from track-finding/export_torchscript.py.
     
    The TrackConstruction class is responsible for constructing all possible track candidates from a set of superpreclusters.