Our Solution
BlackEye is an AI system that processes NASA's Kepler and TESS datasets to identify and characterize exoplanets through a smart approach. The system analyzes 17,232 space objects using 12 basic features plus 3 physics-based smart calculated features, then uses AI to find planets and predict their physical properties.
The Challenge We Address
NASA's Kepler and TESS missions generate vast amounts of light curve data that require smart analysis. Current exoplanet detection relies heavily on manual review by expert astronomers, limiting the pace of discovery and scientific advancement. BlackEye addresses this by automating the analysis of vast amounts of light curve data from NASA's Kepler and TESS missions, reducing manual review time and improving planet detection accuracy. Learn more about the challenges →
How Our AI System Works
BlackEye employs a smart approach that combines planet detection and property prediction to provide comprehensive exoplanet analysis using AI and advanced smart calculations. The system uses AI for gradient boosting, scikit-learn, pandas, and NumPy for AI and data processing, with FastAPI for the Python backend and Docker containerization for deployment.
Smart Approach
Our system performs both planet detection (CONFIRMED/CANDIDATE/FALSE_POSITIVE/UNKNOWN) and property prediction simultaneously using AI models, providing comprehensive exoplanet characterization in a unified framework. The system analyzes 17,232 space objects using 12 basic features plus 3 physics-based smart calculated features, then uses AI to find planets and predict 7 key scientific measurements.
Approach Selection
- AI Classifier: Exoplanet disposition detection (CONFIRMED/CANDIDATE/FALSE_POSITIVE/UNKNOWN)
- AI Regressors: Planet and stellar properties prediction (7 key measurements)
- Smart Calculations: Physics-based features (period_to_duration_ratio, depth_to_mag_ratio, error_quality_score)
- Data Quality Control: Automatic filtering and preprocessing with quality assessment
- Customization Options: Configurable AI parameters for all models
Data Processing Pipeline
- Data Ingestion: Load and standardize multiple astronomical datasets (Kepler KOI, TESS TOI)
- Smart Calculations: Create unified feature set across all surveys (12 basic + 3 smart calculated)
- Classification Training: Train AI classifier on labeled survey data
- Regression Training: Train individual property predictors on confirmed exoplanets
- Unified Prediction: Single interface for both planet detection and property estimation
How We Built BlackEye
Our development process involved extensive research, data analysis, and iterative model refinement to create a robust and accurate exoplanet detection system.
Data Collection & Analysis
Collected and analyzed 17,232 objects from NASA's Kepler and TESS missions, identifying key features and patterns in exoplanet light curves.
Smart Calculations
Developed comprehensive feature extraction methods to capture essential characteristics of planetary transits and stellar variability.
Model Development
Implemented and tested multiple AI algorithms, selecting optimal combinations for planet detection and property prediction tasks.
Validation & Optimization
Conducted extensive cross-validation and customization options to optimize AI models for smart approach.
Platform Integration
Developed RESTful API and web interface to make the technology accessible to researchers and astronomers worldwide.
Special Features & Innovations
BlackEye incorporates several unique features that set it apart from traditional exoplanet analysis tools.
Multi-Mission Integration
Unlike traditional tools that focus on single missions, BlackEye integrates data from multiple NASA missions (Kepler, TESS) into a unified analysis platform, providing comprehensive insights across different observational techniques and time periods.
Real-Time Processing
With efficient AI processing, BlackEye enables real-time analysis of astronomical data, making it suitable for live mission support and rapid response to new discoveries as they happen.
Comprehensive Property Prediction
Beyond simple planet detection, BlackEye predicts 7 key scientific measurements including planet radius, equilibrium temperature, insolation flux, stellar temperature, stellar surface gravity, stellar radius, and signal-to-noise ratio.
Accessibility & Usability
From professional astronomers to researchers, BlackEye's intuitive interface and comprehensive documentation make advanced exoplanet analysis accessible to users with varying technical backgrounds.
Model Customization
Advanced customization options for all AI models (classifier, planet regressor, stellar regressor, quality regressor) with customizable measurements including n_estimators, max_depth, learning_rate, and subsample settings.
Data Split Configuration
Configurable data split ratios for training (70%), validation (20%), testing (5%), and future testing (5%), allowing researchers to customize the model training process for their specific research needs.
Performance Metrics & Technical Specifications
BlackEye delivers exceptional performance across all key metrics, making it the ideal solution for modern exoplanet research and analysis.
Model Performance
- Objects Analyzed: 17,232
- Features Used: 12 raw + 3 engineered
- Parameters Predicted: 7
- Dataset Size: 17,232 objects
- Kepler Objects: 9,564
- TESS TOI Objects: 7,668
Data Sources
- NASA Kepler Objects of Interest (KOI) Dataset
- NASA TESS Objects of Interest (TOI) Dataset
- Supervised Learning with Disposition Columns
API Capabilities
- RESTful API Interface
- Batch Processing Support
- Real-time Analysis
- Custom Model Training
Predicted Measurements
BlackEye predicts 7 key scientific measurements that provide comprehensive characterization of exoplanetary systems.
Planetary Properties
- Planet Radius: Physical size of the exoplanet
- Equilibrium Temperature: Theoretical surface temperature
- Insolation Flux: Stellar radiation received
Stellar Properties
- Stellar Temperature: Effective temperature of the host star
- Stellar Surface Gravity: Surface gravitational acceleration
- Stellar Radius: Physical size of the host star
Quality Metrics
- Signal-to-Noise Ratio: Measurement quality indicator
- Confidence Score: Prediction reliability assessment
- Classification Probability: Category assignment confidence
Ready to Explore the Universe?
Join thousands of researchers and astronomers who are already using BlackEye to accelerate their universe observation process.