Fig. 6From: DeepSeedling: deep convolutional network and Kalman filter for plant seedling detection and counting in the fieldAn example of seedling tracking errors due to inaccurate detection results. The detection model was \(model_{SAll}\) and the testing video was from TAMU2015 dataset. Detection results were inaccurate in Frame 61 and Frame 62, causing incorrect termination of existing trackers and creation of excessive trackers for some seedlings. These tracking errors eventually led to inaccurate counting resultsBack to article page