Deep Learning with Curiosity and Perseverance

In January 2021, the Mars rover Curiosity reached the astonishing age of 3,000 Martian days (sols). While the new rover Perseverance landed on Mars a few months ago, Curiosity continues to capture images of the planet: several hundred thousand camera images are openly available on NASA’s website (https://mars.nasa.gov/msl/multimedia/raw-images/). Such a dataset is ideal for performing various deep-learning tasks that can bring significant value to the Mars exploration community.

Goals

The goal of this project is to develop and implement a set of deep learning algorithms based on unsupervised machine learning. The project has the specific goals of facilitating image/feature search and performing image enhancement (e.g., increasing the resolution of Curiosity’s image set based on the latest and higher resolution images from Perseverance) using state-of-the-art methods.

Scope of work

The student performs the following tasks:

  • Set up project management tools
  • Conduct literature research
  • Create a list of realistic use cases for the project
  • Prepare / clean up the data set
  • Development and implementation of several algorithms addressing the use cases
  • Assess result accuracy
  • Document the work in the form of a scientific report and provide a 1-minute video summarizing the work.

Organization

  • SCS provides the necessary infrastructure for the project (PC, GPUs, desk, etc.)
  • There are weekly meetings between the student and the supervisor
  • The work shall be documented on an ongoing basis
  • The scientific report can be completed and submitted at the end of the project and before the presentation
  • The progress of the work shall be regularly compared with the project plan. Unforeseen problems may require adjustments to the project plan and shall be documented
  • At the end of the project, the computer account must be cleaned up. Only the relevant files should be saved, such as source code, schematics, layouts, configurations and special executable files. Possible follow-up work must be able to start from these files.
  • The student presents his or her work at the end of the thesis at a Tuesday presentation at SCS

Technology

  • Image Processing, Deep Learning
  • PyTorch / Tensorflow
  • Python

Figure Source
Images courtesy of NASA/JPL-Caltech.

Kind of Work
40% theory, 60% engineering

Requirements
First experience with image processing and machine learning.

Time & Effort
Term Paper, 1 Student

Contact
alexis.guanella@scs.ch

Schreibe einen Kommentar

Deine E-Mail-Adresse wird nicht veröffentlicht. Erforderliche Felder sind mit * markiert.