Automated Yield Estimation Of Fruits Using Multilayer Convolution Neural Networks

The yield estimation in agricultural field plays a vital role for the better utilization of resources and to enhance total yield per unit area within the time. The yield estimation of fruit is currently computed manually which leads to labour extensive and also expensive. It also takes much time for yield estimation. This manual sampling may also results in imprecise yield calculation. This makes the demand in machine vision based systems to address the above mentioned problem in detecting the yield estimation and thus reduces the error in counting the number of fruits on each tree. The multilayer CNN proposed here is used to classify the fruit from the tree image and also gives better result even in case of diminished images. Overlapped fruits also counted separately to give better result.