We have added such option in order to make calculated statistics more precise. When image's boundary intersect particles this makes harder to recognize the particles with neural network (because they are not circles anymore). And moreover average particles area value becomes lower than it should be. In order to get this round we exclude all particles whose bounding boxes' centers are less than R pixels near the border of the image. R is calculated as average diameter of all found particles divided by 2 and rounded up.
You can turn this option on when you are starting data processing. And we recommend to do this.
When it is on you can still see excluded particles at the output image. They are marked with pink.
They are still saved in output CSV files but are marked with
1 in the
is_at_border column. As usual, the final statistics you generally need is printed below
-------------------- to the right of the final image.