Computer Science Research and Its Development Journal

Digital Image Classification of Herbal Leaves Using Support Vector Machine and Convolutional Neural Network with Fourier Descriptor Features

Tue, 20 Feb 2024 00:00:00 +0000

Leaves are one component of plants that contain natural properties and are useful for maintaining human health. However, several types of leaves have the same characteristics and characteristics that make it difficult to distinguish. This study aims to classify types of herbal leaves using the SVM method with four kernels (Linear, RBF, Polynomial, Sigmoid) and CNN with Fourier descriptor (FD) feature extraction. The processed dataset is katuk leaf images, and Moringa leaf images of 480 images which are divided into 80% training data and 20% testing data using two scenarios, namely dark and light. From the testing process, it was found that FD + CNN in the light and dark scenarios obtained an accuracy value of 98%. Thus, the FD + SVM algorithm with Linear, RBF, polynomial kernels can be recommended in classifying herbal leaf images to have the best accuracy value of 100%.

IMPLEMENTATION OF DATA MINING CLASSIFICATION FOR DETERMINING THE TYPE OF SOCIAL ASSISTANCE USING THE NAÏVE BAYES CLASSIFIER METHOD

Fri, 08 Mar 2024 00:00:00 +0000

The social assistance program is a program held by the government as an effort to overcome poverty. Mekarjaya Village is one of the villages running the program. In carrying out this social assistance process, there are obstacles in terms of collecting data on its citizens because there are often discrepancies in the recipient data collected by the community with the type of assistance. To make it easier to determine the appropriate type of social assistance, an analysis of the data on the recipients of the social assistance is needed. The data analysis method in this research uses Data Mining including Data Selection and Preprocessing, while the classification method uses the Naïve Bayes Classifier. Testing using the Confusion Matrix produces an accuracy of 94.53% with a comparison of training data and testing 80:20. With this model, it is hoped that village officials can determine the type of social assistance that is appropriate for the community.

An ARTIFICIAL NEURAL NETWORKS USING LEARNING VECTOR QUANTIZATION (LVQ) FOR LEAF CLASSIFICATION

Soeheri, Rita Sari, Wahyu Saptha Negoro, Yuhandri — Fri, 08 Mar 2024 00:00:00 +0000

Leaves are one part of a plant species that is commonly used to classify plant and plant species. The process of assisting various types of leaves usually involves experts using a herbarium, which is a collection of preserved plant specimens. Leaf classification is the detection of different types of leaves, where there are 2 types of leaves including Magnolia Soulangeana and Invillea leaves. The training data contains 30 images consisting of 15 each of the 2 types of leaves, then the test data contains 20 images which are also taken from the 2 types of leaves. So that the total images used are 50 leaf images. The leaf classification uses feature extraction and the method used in the classifier is Learning Vector Quantization (LVQ) which is a pattern classification method in which each output unit represents a particular class or group. The test results showed that the process of calling Magnolia Soulangeana and Bougainvillea leaves in this experiment was successful with 80% detection

Keywords—Leaf classification, Learning Vector Quantization, Artificial Neural Networks, Feature extraction.

Rancang Bangun Saron Elektrik Berbasis Mikrokontroler Arduino Uno CH340 Dengan Sensor Piezoelectric

Dani Rohpandi, Evi Dewi Sri Mulyani, Kiki Abduloh — Fri, 08 Mar 2024 00:00:00 +0000

Gamelan merupakan alat musik tradisional yang cara memainkannya bermacam-macam yaitu dengan cara diketuk atau dipukul, digesek dan ada juga yang dipetik. Dengan adanya teknologi maka banyak pengembangan alat musik yang dibuat, akan tetapi dari beberapa aplikasi alat musik yang dibuat cara memainkannya belum sesuai dengan alat musik yang aslinya, sehingga akan menghilangkan suatu unsur dan nilai dari alat musik tersebut. Salah satu contoh yaitu aplikasi Saron Virtual Instrument berbasis android yang cara memainkannya hanya dengan disentuh oleh jari, sedangkan saron yang aslinya dimainkan dengan cara diketuk atau dipukul. Dengan mengacu pada metode R&D (Research & Development), penulis bermaksud membuat rancang bangun saron elektrik menggunakan mikrokontroler Arduino UNO dan sensor Piezoelectric. Maka hasil akhirnya yaitu dapat membuat saron elektrik yang cara memainkannya sesuai dengan aslinya diketuk sehingga nilai dari alat musik tradisional jenis saron tersebut tetap ada

SISTEM PAKAR DIAGNOSA PENYAKIT PADA SAPI BERBASIS WEB MENGGUNAKAN METODE FORWARD CHAINING

Fri, 08 Mar 2024 00:00:00 +0000

Beternak sapi merupakan bisnis yang memiliki potensi ekonomi yang sangat menjanjikan, namun tingginya permintaan daging sapi dan air susu sapi tidak disertai dengan laju pertumbuhan ternak. Selain itu, kendala yang sering dialami oleh para peternak adalah proses merawat sapi agar terhindar dari penyakit berbahaya dan menular dengan cepat yang dapat berakibat pada kematian. Untuk mencegah agar sapi tidak sakit, maka pemilik sapi harus senantiasa berkonsultasi dengan dokter hewan agar dapat dilakukan pencegahan dan pengobatan terhadap hewan sapi sedini mungkin, namun terbatasnya pakar dan tingginya biaya konsultasi menjadi kendala utama bagi para peternak. Aplikasi pakar ini dirancang sebagai solusi dari kendala yang dihadapi oleh peternak, agar para peternak dapat melakukan konsultasi mengenai penyakit sapi sehingga peternak dapat melakukan penanganan sedini mungkin dari diagnosis yang dihasilkan. Dengan menggunakan metode Forward Chaining, proses pengumpulan fakta dimulai dari gejala yang ditemukan sampai menghasilkan diagnosis sebagai konklusinya dan dengan metode Certainty Factor, hasil diagnosis tersebut diberi nilai persentase atau tingkat keyakinannya

Analysis Of Correlation On Electronic-Based Government System Index Values On The Number Of Web-Defacement Cases Using Linear Regression

Agung Nugroho, Achmad Farid Wadjdi, Teddy Mantoro — Fri, 08 Mar 2024 00:00:00 +0000

The government has established an electronic-based government system (SPBE) program to realize clean, effective, transparent and accountable government governance as well as quality and trustworthy public services implemented with the principles of effectiveness, integration, continuity, efficiency, accountability, interoperability and security. . The implementation of SPBE has been evaluated with the achievement of 16 government agencies receiving a very good title in monitoring and evaluating the implementation of the Electronic Based Government System (SPBE) in 2022. During 2022, there were several agencies with excellent predicates who reported cases of web defacement in their electronic systems. via the zone-h.org site as an open publication media regarding web defacement cases. The National Cyber and Crypto Agency (BSSN) said that during 2022 there would be 2,348 cases of web defacement with the sector most affected by web defacement attacks being the Government Administration sector with a total of 885 cases. This research analyzes the influence between the SPBE index value as the independent variable and the number of web defacement cases as the dependent variable using a simple linear regression statistical method to see the gap in the relationship between the SPBE index and cyber attack cases in the form of web defacement cases that occurred at SPBE agencies. The research was carried out using quantitative research methods to process research data using a simple linear regression method. In the research, the results showed that F count = 2,363978 < from F table = 4,60011 , so it was concluded that there was no relationship between the SPBE index value and the number of web defacement cases in agencies that received excellent predicate

Smart Lock Application for Smart Security System in Cargo Shipment

Faisal Lubis — Fri, 01 Mar 2024 00:00:00 +0000

Transportation services for cargo shipment is growing rapidly during the pandemic and the post pandemic of Covid-19. However, this business is not without problems. Security disturbances during the cargo shipment process are a major problem, and this often happens in Indonesia. For this reason, this research will build a smart security system using automation technology and based on the Internet of Things (IoT). The system is a Smart Lock installed on the cargo. This smart lock can be monitored remotely, such as opening or closing the lock and the condition of the cargo in the field. Only the cargo owner can open or close the cargo lock. In addition, this system is also equipped with temperature and humidity sensor to maintain the quality of the goods in the cargo. Cargo position, temperature and temperature data can be monitored via the website by the cargo owner. The results of this research will be useful for the security system and convenience of the cargo delivery process and the digitization process will be easier to do in the future.

Comparison of Waste Type Classification Using Convolutional Neural Network with ResNet18 and ResNet50 Architecture

Christin Evasari Nainggolan, Muhammad Nasir, Fatoni, Devi Udariansyah — Tue, 19 Mar 2024 00:00:00 +0000

Complex problems occur in dealing with waste, both in developing and developed countries, such as Indonesia. According to data from the Ministry of Environment and Forestry, in 2022 the total waste pile will reach 34,439,338.12 tons per year. In this research, machine learning will be used by comparing CNN architecture, ResNet18 with ResNet50, for the classification of waste types. This research uses 2527 images of garbage image data consisting of 6 classes, namely cardboard, glass, metal, paper, plastic and trash. Convolutional Neural Network is a component of the Deep Neural Network method that has the ability to identify objects in images with a high level of complexity. From the ResNet18 model test in this study, the accuracy was 98.69% and the test results on ResNet50 resulted in an accuracy of 99.41%. The precision and recall results of both models reflect excellent performance and accuracy of around 99%. So it can be concluded that both CNN models, ResNet18 and ResNet50, have excellent performance in classifying garbage images.