Hunt for distant worlds using real NASA Kepler data! This neural network pipeline automatically detects exoplanet transits from light curves, bringing professional planet discovery to your desktop.

Every time a planet crosses in front of its star, it creates a telltale dimming signature in the star’s light curve—a cosmic fingerprint that reveals worlds orbiting distant suns. This Global Nominee project tackles one of astronomy’s most exciting challenges: automatically detecting these subtle transit events from the vast treasure trove of NASA Kepler mission data.

Built with TensorFlow and the Lightkurve library, this system creates a complete machine learning pipeline that downloads real Kepler light curves, processes time-series flux data into 500-point segments, and trains a neural network classifier to distinguish transit signals from stellar noise. The architecture features dropout regularization and handles the complexities of real astronomical data, while intelligently falling back to synthetic datasets for offline development. It outputs trained models, cached datasets, and visualization tools that make exoplanet detection accessible to researchers and citizen scientists alike.

Whether you’re a computational astronomer looking to streamline transit detection workflows or a developer fascinated by space data, this toolkit demonstrates how modern AI can accelerate the discovery of new worlds. The project bridges professional astronomical research with practical machine learning implementation, offering both educational value and real scientific utility.

⭐ Stars: 3
💻 Language: Python
🔗 Repository: Skywalker690/nasa-space-apps