Hello, I am using tiny yolov2 algorithm with my custom object recognition and tracking. But
I have one problem when I add new class on labels.txt at the same time added new photo my images path and xml path. When accuracy of classes is hardly decreased. İmages of all classes is various like first class has 1.450 photo and second class 473 photo etc... All classes have different photo. Due to This situation would I encounter an error.? Because ı have error about when class number increased my accuracy of each classes is decreased as hardly. such as %50-60 threshold will be down %5-10. Can you help to me about this subject. What can I do ? Thank you.
Hello, I am using tiny yolov2 algorithm with my custom object recognition and tracking. But
I have one problem when I add new class on labels.txt at the same time added new photo my images path and xml path. When accuracy of classes is hardly decreased. İmages of all classes is various like first class has 1.450 photo and second class 473 photo etc... All classes have different photo. Due to This situation would I encounter an error.? Because ı have error about when class number increased my accuracy of each classes is decreased as hardly. such as %50-60 threshold will be down %5-10. Can you help to me about this subject. What can I do ? Thank you.