Circle detection for FARICH
Rings detection for FARICH
Status: This repository is not supported
We have trained nn to detect rings with this tutorial.
Dataset contains 100x100px boards with one ring on each and uniform noise 1%.
Technical details you can see in the notebook
Precision and recall vs treshold predictor score:
Some detection examples
Good/perfect examples:
Bad example:
Most problems from such images but even we cannot detect there rings. Why should nn?
We also trained nn for many circles (see notebook)
Precision and recall vs treshold predictor score: