Fig. 1From: DeepSeedling: deep convolutional network and Kalman filter for plant seedling detection and counting in the fieldCotton plant seedlings and weeds detected in representative images of the \(Seedling_{All}\) testing set by the Faster RCNN model that was trained using the \(Seedling_{All}\) training set. a, b Are images originally collected in the TAMU2015 dataset, presenting challenges of high seedling occlusion; c, d in the UGA2015 dataset, presenting challenges of extreme illumination; and e, f in the UGA2018 dataset without critical challenges for object detectionBack to article page