How aditi012345/packetpulse works
SentinelNet is a Streamlit network-intrusion detection dashboard with two user workflows: score uploaded traffic data using a selected dataset/model, or capture live IP traffic and score reconstructed flows with the CICIDS XGBoost model. The live path extracts bidirectional flow features before inference; both paths use shared model-loading and prediction helpers. The graph also shows the saved model artifacts. The README describes the NIDS goal, but sampled files do not establish training or deployment workflows beyond these scoring paths.
Main entry points: live_flow_predictor.py Live Capture Scoring · app.py Streamlit Dashboard · model_loader.py Model Loader · predictions.py Prediction Helper
Components 10
Dashboard
- Streamlit Dashboard
app.py - CSV Scoring
app.py - Live Capture Scoring
utils/live_flow_predictor.py
Traffic Analysis
- Flow Feature Extraction
utils/cicids_flow_features.py - Threat Verdicts
utils/live_flow_predictor.py
Models
- Model Loader
utils/model_loader.py - Prediction Helper
utils/predictions.py - Saved Models
models
Other components
- Analyst
- IP Traffic
Connections 11
- Analyst → Streamlit Dashboard: selects mode
app.py - Streamlit Dashboard → CSV Scoring: routes CSV scoring
app.py - Streamlit Dashboard → Live Capture Scoring: starts capture
app.py - IP Traffic → Live Capture Scoring: supplies packets
utils/live_flow_predictor.py - Live Capture Scoring → Flow Feature Extraction: extracts features
utils/live_flow_predictor.py - Live Capture Scoring → Model Loader: loads model
utils/live_flow_predictor.py - Live Capture Scoring → Prediction Helper: predicts flows
utils/live_flow_predictor.py - Live Capture Scoring → Threat Verdicts: builds verdicts
utils/live_flow_predictor.py - CSV Scoring → Model Loader: loads model
app.py - CSV Scoring → Prediction Helper: predicts rows
app.py - Model Loader → Saved Models: reads pickle
utils/model_loader.py