Skip to main content

Table 3 Regression results between seedling counts obtained by the proposed method and human field assessment

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

Detection modelTesting videosRegression equationVideo quantity\(R^{2}\)RMSEMAEMRE (%)
\(model_{SAll}\)AllY = X750.981.61.57
\(model_{SAll}\)TAMU2015Y = 0.96 X250.853.33.411
\(model_{SAll}\)UGA2018Y = X250.990.20.56
\(model_{SAll}\)UGA2015Y = 1.01 X250.961.10.84
\(model_{U15U18}\)TAMU2015Y = 0.95X250.803.93.411
\(model_{U15T15}\)UGA2018Y = 1.02X250.680.80.67
\(model_{T15U18}\)UGA2015Y = 0.32X + 18.65250.334.312.157
  1. \(R^{2}\) refers to adjusted \(R^{2}\)
  2. RMSE root mean squared error, MAE mean absolute error, MRE mean relative error