Wouldn’t it be great if the picture would already take the hue, saturation and magic “value” settings into account and show the filtered image? Thus one could clearly see why it detects some areas faulty…?
But it then takes this rotation into account and cuts the image. Does it dewarp the image as well?
What is the convention, how does it rotate an align the image?
Today I tried a little bit more and this confuses me. It triggered correctly the two red dots, as I understand from your explanations and this picture here:
but it says in the log:
How can I see those 18 objects, where are they? The message does not correlate with the image I see above.
If I then run weed detection (just random on the same image) and press test, it overlays the coordinates (I guess?)
like this:
But they are wrong. How are they calculated what does that take into account? I guess if they are not matching between coordinate where the FB is vs. weed detected the system is not able to decide wether its a weed or a seedling, right?
Please give some more text in the wiki as well…
Thanks very lot for the fast responses, @Gabriel! That is needed and appreciated!