Hi, we have a network that has been trained in a particular way, and sometimes it hasn't a great accuracy. We noticed that, for the mAP calculation in instance segmentation, the script requires at least one prediction per frame. However, there are frames where we fail to detect any instance. How can we run an evaulation on those? Thanks in advance.
Hi, we have a network that has been trained in a particular way, and sometimes it hasn't a great accuracy. We noticed that, for the mAP calculation in instance segmentation, the script requires at least one prediction per frame. However, there are frames where we fail to detect any instance. How can we run an evaulation on those? Thanks in advance.