Anomaly Detection on Mars Using Deep Learning
  • The Results

Anomaly Detection on Mars Using Deep Learning

Project completed!

38 Collaborators built an anomaly detection model for identifying past or present extraterrestrial technology on the surface of Mars. A U-Net model yielded the best scores with precision measures for all anomalies of above 90 percent.

 

Why applying anomaly detection on Mars

The second-smallest planet in the Solar System comprises a thin atmosphere and has surface features reminiscent both of the impact craters of the Moon and the valleys, deserts, and polar ice caps of Earth.

Recently looking for extraterrestrials in the form of technosignatures has gained new interest. These signatures are measurable properties that provide scientific evidence of past or present extraterrestrial technology. Scientists want to evaluate how far the search for technosignatures has come and what the most promising possibilities for the future are.

 

The solutions

Below is a brief video demonstrating how the community worked together including the end results.

Among the many tasks accomplished, the team applied GAN`s for building an expert system that classifies images with an anomaly score. A Python package to process data from Mars efficiently. And lastly, testing various models to identify the best-fit model, which turned out to be the U-Net model.

 

Feedback from one of the collaborators:

In a world being plagued by greed, hate, and intolerance, Omdena comes as a breath of fresh air to do away with national barriers. This project is a testament to the fact that bringing together a group of strangers from different corners of the Earth, who have never met each other before; transcending geographical borders and time zones to work together and solve fascinating social problems; whilst learning from and inspiring each other every single day, is not just a pipe dream, thanks to online education, collaborative tools and platforms like Omdena!

 

 

This project has been hosted with our friends at Univ. of Bern, Switzerland

Articles from the project