Nice · 43°43′ N · 7°18′ E

← Back to portfolio

VF·22 · Project catalogue

SExtractor driven by Machine Learning

Attempt to design an object detection machine learning model to detect light sources in astronomical images

Designation
VF·22
Period
Fields
Machine LearningAstroAcademic
Link
View project →
AIThis image was generated by AI solely to illustrate the project, it is not part of the project itself.
SExtractor driven by Machine Learning
AIThis image was generated by AI solely to illustrate the project, it is not part of the project itself.
SExtractor driven by Machine Learning

Having an algorithm that allows detecting light sources in an image is obviously crucial in astronomy. But such algorithms are heavy to run and give a positive or negative result, no matter if it is in the gray zone. So, what if we use an object detection model to perform such tasks? It can probably run faster and give a kind of uncertainty on each detection.

The goal of this project was to try to make such an ML model. Unfortunately, due to the unknown number of outputs, designing such a model was much more complex than initially thought and it was not possible to create it in 20 hours (officially, but near 50 hours in practice) without a deep mastering of the ML world.

Status

Completed

Organization

UBFC, Besançon, France