Home
Intelligent SETI Signal Analysis — Rust-Accelerated Processing with Machine Learning Classification
MitraSETI is a ground-up reimagination of the SETI signal analysis pipeline, combining a Rust-powered de-Doppler engine (up to 45x faster on real Breakthrough Listen data), a CNN + Transformer classifier that automatically rejects RFI and flags anomalies, and a streaming observation mode for multi-day unattended campaigns — with desktop and web interfaces for real-time monitoring.
| Page | Description |
|---|---|
| Architecture | System overview, data flow diagrams, component interaction, threading model, and state management |
| Pipeline Deep Dive | Stage-by-stage walkthrough of the full processing pipeline from file ingest to result export |
| ML Model Architecture | CNN + Transformer signal classifier, OOD detection ensemble, 9-class taxonomy, and training pipeline |
| Rust Core | The mitraseti-core crate — de-Doppler engine, filterbank reader, RFI filter, and PyO3 bindings |
| Page | Description |
|---|---|
| Benchmark Results | Real-data benchmarks against turboSETI with methodology and analysis |
| Streaming Mode | Continuous observation guide — auto-training, self-correcting thresholds, cadence analysis, daily reports |
| API Reference | Full FastAPI endpoint documentation with curl examples |
| Page | Description |
|---|---|
| Comparison with turboSETI | Feature-by-feature comparison, architectural differences, when to use each tool, and migration guide |
- GitHub Repository: deepfieldlabs/MitraSETI
- License: MIT
- Python: 3.10+ | Rust: 1.70+
- Author: Saman Tabatabaeian · Deep Field Labs
- 45x faster processing on million-channel observations via parallel Rust de-Doppler search
- Two-stage ML classification — rule-based filtering + CNN+Transformer inference
- Out-of-distribution detection — ensemble of MSP, Energy, and Spectral distance methods
- 9-class signal taxonomy — from NARROWBAND_DRIFTING to CANDIDATE_ET
- Streaming observation mode — multi-day campaigns with auto-training and daily HTML reports
- Catalog cross-matching — SIMBAD, NVSS, FIRST, and ATNF Pulsar catalogs
- AstroLens integration — optical + radio cross-reference
- Desktop + Web UI — PyQt5 desktop app and FastAPI web interface with WebSocket live streaming
-
Format support — Sigproc
.filand HDF5.h5(Breakthrough Listen format)