Hello, I have written a programm which can do object detection on every frame with acceptable time and accuracity. Great(Took me tons of time to try various libs)!
Now i want to count the detected objects when they passed a certain line in the image. I think when i as a human count objects - i dont count the twice as i can remember what i counted. A program cannot do this out of the box when i detect them on every frame.
Can open cv help me here via the open cv object tracking api? I am hoping open cv will recognize the object again(is it using template matching?) Am i thinking to complicated and should just look for some overlapping of bounding boxes with the "virtual line" in the image? What if the objects have different speeds?
I want no implementation(ok i would take it - but not my aim ^^) but maybe a comment on this would help. Thx a lot again + Greetings, Holger