Home
AI that watches the sky and finds what others miss.
AstroLens is an open-source astronomical anomaly detection tool that uses deep learning to identify unusual objects in sky survey images -- automatically, at scale, and without human supervision.
 |
Anomaly Detection -- AI identifies unusual objects with OOD scoring and YOLO transient detection |
 |
Gallery & Viewer -- Browse, inspect, and verify detected anomaly candidates |
 |
Web Dashboard -- Statistics, streaming charts, and analysis insights |
- Overview -- What AstroLens does and how it works
- Galaxy Mode -- Vision Transformer + OOD anomaly detection
- Transient Mode -- YOLO object detection for supernovae and transients
- Streaming Discovery -- Autonomous multi-day operation (v1.1.0)
- API Reference -- REST API endpoints
- Troubleshooting -- Common issues and solutions
Version 1.1.0 introduces Streaming Discovery -- a fully autonomous pipeline that downloads images from major astronomical archives, classifies them using a Vision Transformer and YOLO transient detector, self-corrects its detection thresholds in real time, and cross-references every finding against the SIMBAD international astronomical database.
Autonomous validation run results:
- 22,195 images analyzed from 7 survey sources across 5,471 unique sky regions
- 3,541 anomaly candidates identified
- 269 known objects independently recovered and confirmed against SIMBAD/NED
- YOLO transient detection: 99.5% mAP50
- Notable recoveries include SN 2014J, NGC 3690, SDSS J0252+0039