Neural Visual Detection of Grain Weevil (Sitophilus granarius L.)

Abstract

A significant part of cereal production is intended for agri-food processing, which implies a necessity to search for and implement modern storage systems for this product. Stored grain is exposed to many unfavorable factors, particularly caryopsis macro-damage caused mainly by grain weevil (Sitophilus granarius L.). This triggers a substantial decrease in the value of the stored material, thus resulting in serious economic losses. Due to this fact, it is necessary to take steps to effectively detect this pest’s presence when grain is delivered to storage facilities. The purpose of this work was to identify the representative physical characteristics of wheat caryopsis affected by grain weevil. An automated visual system was developed to ease the detection of damaged kernels and adult weevils. In order to obtain the empirical data, a decision was made to take advance of SKCS 4100 (the Perten Single Kernel Characterization System). The measurements obtained were used to build the training sets necessary in the process of ANN (artificial neural network) learning with digital neural classifiers. Next, a set of identifying neural models was created and verified, and then the optimal topology was selected. The utilitarian goal of the research was to support the decision-making process taking place during grain storage.

Quality Evaluation of Potato Tubers Using Neural Image Analysis Method

“…Those techniques allow for a non-invasive inside view and provide additional information on the analyzed object. During the research it was proved that these methods are suitable for accurate detection in all tested classification models [21,22].…”

Definition and Application of a Computational Parameter for the Quantitative Production of Hydroponic Tomatoes Based on Artificial Neural Networks and Digital Image Processing

“…For identification and classification purposes, this work uses a specially designed and trained neural network. The virtues of this computational technique are well known in applications of machine learning [33,34]. In Figure 2, the general outline of the neural network model for a pixel classification process is presented.…”

Artificial neural network approach for predicting the sesame (Sesamum indicum L.) leaf area: A non-destructive and accurate method

“…Examples include variety identification in seeds [ 31 ] and in intact plants by using leaves [ 32 ]. The ANNs were applied in research with the identification of grain pests [ 33 ], modeling of dynamic responses of plant growth affected by climate change [ 34 ], estimation of corn grain yield [ 35 ], estimation of soil quality for crops [ 36 ], prediction of the area of harvest, yield, and production [ 37 ], prediction of greenhouse gas emissions [ 38 ], forecasts on the accumulation of heavy metals in crops [ 39 ] and determine seed germination [ 40 ]. In addition, non-destructive predictive models using ANN to estimate leaf area were proposed for species such as durian [ 13 ], cabbage [ 41 ], wedelia [ 42 ], tomato [ 43 ] and corn [ 44 ].…”