Atmospheric cloud boundary detection using machine learning
Rusen Oktem, Project Scientist
Earth and Planetary Science
Closed. This professor is continuing with Fall 2023 apprentices on this project; no new apprentices needed for Spring 2024.
We have been collecting camera images at multiple locations to study atmospheric clouds for the past 6+ years. Extracting clouds from the background is an essential and challenging step before processing these images. The challenges mostly arise from varying environmental conditions and cloud characteristics. The expected outcome of the project is to develop and implement a python code that will identify and extract atmospheric cloud boundaries from the background using neural networks.
Role: The project tasks are:
- Sorting of a large archive of cloud images according to cloud types.
- Generating a training set of images by labeling cloud vs background boundaries.
- Developing a NN algorithm and testing its performance against the training set.
Qualifications: At least basic experience with python code development and use of NN packages is a must. Basic understanding of image processing and image data handling is a plus.
Hours: 3-5 hrs
Mathematical and Physical Sciences