Skip to main content
Fig. 5 | Plant Methods

Fig. 5

From: DeepSeedling: deep convolutional network and Kalman filter for plant seedling detection and counting in the field

Fig. 5

Regression results between seedlings counts calculated by the developed approach and human field assessment. a Results obtained using the \(model_{SAll}\) detection model for all testing videos (n = 75, lifetime threshold of 7 was used for the TAMU2015 and UGA2018 testing videos and 15 for the UGA2015 testing videos). b Counting differences between the proposed method and field assessment. A total of 53 videos (70%) had the counting differences less than 1 seedling, and 68 videos (90%) had the counting differences less than 5 seedlings

Back to article page