<span>Presently, the demands for rice are increasing. This will affects the need for producing and sorting rice grain in faster and exceed the normal requirement. However, the manual rice classification using naked eyes are not very accurate and only professionals are able to do it. Machine learning is found to be a suitable technique for rice classification in producing an accurate result and faster solution. Thus, a study on the classification of rice grain using an image processing technique is presented. The rice grain image went through the pre-processing process which includes the grayscale and binary conversion, and segmentation before the feature extraction process. Four attributes of shape descriptor which are area, perimeter, major axis length, and minor axis length and three attributes of color descriptor which are hue, saturation and value were extracted from each rice grain image. In another note, a Multi-class Support Vector Machine (SVM) is used to classify the three types of rice grain which are basmathi, ponni and brown rice. The performance of the proposed study is evaluated to 90 testing images which returned 92.22% of classification accuracy. The study is expected to assist the Agrotechnology industry in automatic classification of rice grain in the future.</span>

<span lang="EN-US">Maize is one of the world's leading food supplies. Therefore, the crop's production must continue to reproduce to fulfill the market demand. Maize is an active feeder, therefore, it need to be adequately supplied with nutrients. The healthy plants will be in deep green color to indicate it consist of adequate nutrient. Current practice to identify the nutrient deficiency on maize leaf is throught a laboratory test. It is time consuming and required agriculture knowledge. Therefore, an image processing approach has been done to improve the laboratory test and eliminate a human error in identification process. The purpose of this research is to help agriculturist, farmers and researchers to identify the type of maize nutrient deficiency to determine an action to be taken. This research using image processing techniques to determine the type of nutrient deficiency that occurs on the plant leaf. A combination of Gray-Level Co-Occurrence Matrix (GLCM), hu-histogram and color histogram has been used as a parameter for further classification process. Random forest technique was used as classifiers manage to achive 78.35% of accuracy. It shows random forest is a suitable classifier for nutrient deficiency detection in maize leaves. More machine learning algorithm will be tested to increase current accuracy.</span>

The Artificial Neural Network (ANN) is an Artificial Intelligence technique which has the ability to learn from experiences, enhancing its performance by adapting to the environmental changes. The key benefits of neural networks are the prospect of processing vast quantities of data effectively, and their ability to generalize outcomes. Considering the great potential of this technique, this paper aims to establish a performance evaluation of Multilayer Perceptron (MLP) and a Radial Basis Function (RBF) networks in investigating the contributing factors for COVID-19spread and death. The RBF and MLP networks are typically used in the same form of applications, however, their internal calculation structures are different. A comparison was made by using a dataset of COVID-19 cases in 41 Asia countries during April 2020. There are nine contributing factors which acted as the covariates to the network such as Cases, Deaths, High Temperature, Low Temperature, Population, Percentage of Cases over Population, and Percentage of Death over Population, Average Temperature, and Total Cases. The results obtained from the testing sets indicated that the two neural structures were able to investigate the COVID-19 spread and death contributing factors. Nevertheless, the RBFnetwork indicated a slightly better performance than the MLP.

White blood cells (WBCs) are significant element in the immune system to shield against infections. The health condition of a person can be determined from the WBCs as it functions to produce and react to illnesses. However, there are challenges in processing a massive amount of blood samples due to time constraint and skills, which limit the speed and accuracy of classifying the WBCs. Thus, this paper conducts a comparative analysis of Support Vector Machine (SVM) and Convolutional Neural Network (CNN) techniques for WBCs classification. A process of feature extraction is performed to analyze the characteristics of WBCs by extracting the colour, texture, and shape. The classification performance of each technique is tested to 200 of WBCs images. The classification of the WBCs is divided into five different types of neutrophil, basophil, eosinophil, lymphocyte, and monocyte. Upon the testing conducted, the SVM reflected 88.5% of classification accuracy, whereas the CNN on the other hand returned a higher percentage of 94%. Thus, it is proven that CNN is observed to return a better WBCs classification outcome as compared to the SVM.

No abstract