Rice Seed Varieties Identification based on Extracted Colour Features using Image Processing and Artificial Neural Network (ANN)

Abstract

Abstract-Determination of rice seed varieties is very important to ensure varietal purity in the production of high-quality seed. To date, manual seed inspection is carried out to separate foreign rice seed varieties in rice seed sample in the laboratory as there is lack of an automatic seed classification system. This paper describes a simple approach of using image processing technique and artificial neural network (ANN) to determine rice seed varieties based on extracted colour features of individual seed images. The experiment was conducted using 200 individual seed images of two Malaysian rice seed varieties namely MR 219 and MR 269. The acquired seed images were processed using a set of image processing procedure to enhance the image quality. Colour feature extraction was carried out to extract the red (R), green (G), blue (B), hue (H), saturation (S), value (V) and intensity (I) levels of the individual seed images. The classification using ANN was carried out by dividing the data sets into training (70% of data), validation (15%) and testing (15%) dataset respectively. The best ANN model to determine the rice seed varieties was developed, and the accuracy levels of the classification results were 67.5% and 76.7% for testing and training data sets using 40 hidden neurons.

Prediction and qualitative analysis of sensory perceptions over temporal vectors using combination of artificial neural networks and fuzzy logic: Validation on Indian cheese (paneer)

“…As a result, R 2 values observed a slight decrease for combined prediction, yet the predictions were good enough with R 2 = 0.97 at 10‐9‐4 (input‐hidden‐output) neuron to be used for further analysis. The decreased R 2 are consistent with increased complexity and variance and reduction in correlation (for validation between experimental data), because the increase in output nodes is normally observed (Aznan et al., 2017; Bisgin et al., 2018).…”

“…These fuzzy scales compare products on a discrete point of time, rather than evaluating and comparing them for acceptance throughout the shelf life (Mukhopadhyay et al., 2013). The mathematical tools, such as artificial neural networks (ANNs) and fuzzy logic, have been extensively used for predicting the product shelf life and comparing products based on sensory parameters, respectively (Aznan et al., 2017). They also address the complex classification problems for characterizing the food items using some initial parameters (Granato et al., 2018).…”

Application of near-infrared hyperspectral imaging to identify a variety of silage maize seeds and common maize seeds

“…It is difficult to classify seeds with similar external shape. 10 Protein electrophoresis technique is based on the content of protein in seeds of different varieties and the speed of protein molecules swimming in the electric eld to identify the variety of seeds. 11 It is accurate and effective, but requires professional operation.…”

PhosopNet: An improved grain localization and classification by image augmentation

“…Many researchers leveraged computer vision as an inspection solution since the beginning of "Japanese rice grading problem" in 2002 [7]. From the literature, all papers could be categorized by methods [8] into 2 groups: bag of words [9][10][11][12][13][14][15][16][17][18][19][20][21][22][23] and convolutional neural network (CNN) [24][25][26][27][28][29]. The former was used by early researches [30] (which require less number of labeled data [31]) that were still useful in some open-world industry [20][21][22] such as iRSVPred [23].…”