Please use this identifier to cite or link to this item:
|Title:||Development and assessment of different modeling approaches for size-mass estimation of mango fruits (Mangifera indica L., cv. 'Nam Dokmai')|
|Keywords:||Agricultural and Biological Sciences|
|Abstract:||© 2015 Elsevier B.V. To meet consumer demands, the need for quick and accurate methods for quality assessment of fresh fruits is increasing constantly. This study presents a comparison of three different models for mass estimation of mango fruits (cv. 'Nam Dokmai') calculated by simple linear regression (SLR), multiple linear regression (MLR) and artificial neural network (ANN). Three dimensions (length, maximum width, and maximum thickness) were manually measured and included as parameters for model building. Calibration and validation were carried out on independent data sets with 820 samples (2010-2012) and 61 samples (2014), respectively. This allowed it to establish a high-performance model that can be used for further mass-size estimation in a machine-vision system. For the SLR, an existing equation for mass estimation was modified to calculate an adjusted coefficient for accurate mass estimation. A MLR model was proposed to obtain the intercept, the slopes of the three parameters length, maximum width and maximum thickness as well as the random error. In addition, an ANN model was used as it allows the network to learn linear and nonlinear relationships between inputs and outputs. Performance evaluation of three different models was based on a compilation of different statistical error parameters and goodness-of-fit measures and the outcomes of the models were compared. ANN was found to be the most accurate and robust model for mass estimation with a root mean squared error (. RMSE) of 6.55. g, mean absolute percentage error (. MAPE) of 1.62%, and coefficient of efficiency (. E) of 0.99 after validation. Therefore, it can be applied for mass estimation of mango fruits with highest accuracy and success rate of 96.7% compared to the other models in this study.|
|Appears in Collections:||CMUL: Journal Articles|
Files in This Item:
There are no files associated with this item.
Items in CMUIR are protected by copyright, with all rights reserved, unless otherwise indicated.