Using the artificial neural network to estimate leaf area

- In this study, leaf areas of 61 plant species with different leaf shapes were estimated by ANNs.
- Results showed that the ANNs could provide good estimation of leaf area.
- Input variable combination affected the ANNs performance to estimate the leaf area.
- MinMax pre-processing method increased accuracy of prediction.

Artificial neural network (ANN) is applied for many subjects in agricultural science such as: crop yield and evapotranspiration prediction, soil parameters estimation, water demand forecasting, hydrological forecasting. Leaf area is one of parameters that is used to assess the plant vegetative growth. In this study, leaf areas of 61 plant species with different leaf shapes were estimated by ANNs and the effect of input data and pre-processing methods on ANNs performance was assessed. Results showed that the ANNs could provide good estimation of leaf area. ANNs input variable combination affected the ANNs performance to estimate the leaf area. With increase in number of hidden layers the epochs decreased and accuracy of the leaf area prediction and running speed increased. Results of test data set showed that MinMax pre-processing method resulted in more accurate prediction in comparison with the no pre-processed method and Norm STD method. The most conclusive result of this study is the application of ANNs for all of plant species, whereas, in application of other methods: specific equation should be prepared for each plant.