Skip to content

BabiDoo/nasa-space-app-2025

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

28 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

🌌 ExoSeeker

Made with Python
Framework PyTorch
License MIT
NASA Data

ExoSeeker is an end-to-end Artificial Intelligence platform designed to accelerate the discovery of exoplanets.
It unifies heterogeneous datasets from NASA missions Kepler, K2, and TESS, applying Explainable AI for robust and transparent classifications.


🎥 Video Presentation

This is the official video presentation of the ExoSeeker project for the NASA Space Apps Challenge:

▶️ Watch the Video Presentation

💻 Project Official Site

This is the official site of the ExoSeeker project for the NASA Space Apps Challenge:

🔧 Visit your Official Site 🚀


📖 Summary

ExoSeeker automates the classification of exoplanet candidates.
We tested multiple models: Decision Tree, KNN, Naive Bayes, Random Forest, LightGBM, and 1D-CNN, prioritizing F1-score as the main metric.

Key features:
✅ Consolidates data from NASA catalogs
✅ Balances and cleans datasets
✅ Uses Explainable AI (SHAP + Grad-CAM)
✅ Provides web interface for analysis & PDF reports
✅ Open-source, reproducible, and globally accessible


🔬 Project Details

ExoSeeker addresses one of astronomy’s greatest challenges:
the exponential growth of exoplanet data that still requires time-consuming human validation.

🛠️ Technical Highlights

  • Languages & Frameworks: Python, PyTorch/Lightning, LightGBM
  • Optimization: Optuna (hyperparameter tuning)
  • Web Stack: FastAPI, Next.js, Plotly
  • Data Storage: PostgreSQL, Redis
  • Reproducibility: conda-lock environments, fixed seeds

⚙️ How It Works

  1. Data Pipeline → Ingests Kepler, K2, and TESS datasets.
  2. Training → Runs multiple ML models for robust classification.
  3. Explainability → SHAP + Grad-CAM show why a signal is labeled as a planet.
  4. Interface → Upload data, run analysis, view metrics, download PDF reports.

🤖 Use of Artificial Intelligence

AI was the core of this project:

  • Classification: ML models detect planets vs. false positives.
  • Evaluation: Balanced datasets, hyperparameter tuning, F1-score optimization.
  • Explainability: SHAP & Grad-CAM for transparency.
  • Integration: Web interface connects models to user workflows.

Acknowledgment of AI Use

  • All AI outputs are documented and labeled.
  • No NASA logos or branding were used.
  • Any illustrative AI-generated media includes watermarks.

🛰️ NASA Data Sources



👩‍🚀 Team Members

Name Country Role
Marina Corrêa Freitas ⭐ (Team Owner) 🇧🇷 Brazil 🚀 Project Lead
Luiza Arievilo 🇧🇷 Brazil 🔬 Research
Márcia Saori Câmara Kishi 🇧🇷 Brazil 🎨 Research & Design
Jannaina Anita Sangaletti 🇧🇷 Brazil 🤖 Machine Learning & Data
Samantha Nunes 🇧🇷 Brazil 💻 Frontend & UX
Barbara Lais Dorneles Martins 🇧🇷 Brazil 🔧 Backend & Integration

ExoSeeker accelerates exoplanet discovery and makes science transparent, reliable, and open to all.

About

ExoSeeker is an end-to-end Artificial Intelligence platform designed to accelerate the discovery of exoplanets. It unifies heterogeneous datasets from NASA missions Kepler, K2, and TESS, applying Explainable AI for robust and transparent classifications.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors