Claim Missing Document
Check
Articles

Found 39 Documents
Search

Optimasi Convolutional Neural Network NASNetLarge Menggunakan Augmentasi Data untuk Klasifikasi Citra Penyakit Daun Padi Afiana Nabilla Zulfa; Jasril Jasril; Muhammad Irsyad; Febi Yanto; Suwanto Sanjaya
JURNAL MEDIA INFORMATIKA BUDIDARMA Vol 7, No 2 (2023): April 2023
Publisher : Universitas Budi Darma

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30865/mib.v7i2.6056

Abstract

Diseases that attack rice are one of the elements that can reduce rice production. Rice diseases include Blast, Brown Spot, Leaf Smut, and so on. Distinguishing rice disease from sight has a weakness because rice disease has similar symptoms and characteristics. Farmers lack knowledge in identifying rice disease types so that technology is needed that can help distinguish rice diseases. The method used for rice image classification in this study is the Convolutional Neural Network NASNetLarge architecture. There are two classification processes, namely the classification process using data augmentation and without data augmentation. The data consists of 4 classes, namely Healthy, Leaf Smut, Blast, and Brown Spot with a total of 440 original images and 1320 augmented images. This study uses data augmentation, namely Horizontal Flips, Vertical Flips, and Contrast. The results for the classification process without data augmentation obtained the highest accuracy, namely 94.31%, 100% precision, 100% recall, and 100% f1-score at a ratio of 80:20, learning rate 0.1, dense 256, batch size 32, and optimizer Adam. While the accuracy obtained in the classification process using data augmentation is 98.73%, 96.11% precision, 100% recall, and 98.01% f1-score at a ratio of 70:30, learning rate 0.1, dense 16, batch size 128, and the Adagrad optimizer. The accuracy results show that the data augmentation and hyperparameters used can increase the accuracy in classifying rice leaf disease images.
Klasifikasi Citra Stroke Menggunakan Augmentasi dan Convolutional Neural Network EfficientNet-B0 Nadila Handayani Putri; Jasril Jasril; Muhammad Irsyad; Surya Agustian; Febi Yanto
JURNAL MEDIA INFORMATIKA BUDIDARMA Vol 7, No 2 (2023): April 2023
Publisher : Universitas Budi Darma

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30865/mib.v7i2.5981

Abstract

A stroke is a sudden onset of brain dysfunction, lasting for 24 hours or longer, resulting from clinically focal and global brain dysfunction. As many as 15 million people die from stroke each year. The stroke patients need an immediate treatment to minimize the risk of brain damage. One of the proponents for the stroke diagnosis is through a computed tomography (CT) image. In recent years, the image processing techniques capable to detect stroke patterns in a brain image, it can be useful for doctors and radiologists in doing diagnosis and treatment. This study aims to compare the level of accuracy using augmentation and without augmentation and hyperparameters using the Convolutional Neural Network in the EfficientNet-B0 architecture to classify ischemic, hemorrhagic, and normal brain stroke images. The data augmentation is produced by rotating, horizontal flipping, and contrast tuning of the original data. Testing data is provided as much as 20% of the portion of the original and augmented data, and the other 80% is used for the training process to find the optimal model. The model search is based on the composition of the training and validation data with a ratio of 70:30, 80:20 and 90:10. The experimental results show that the best performance is obtained for the combined original and augmented images, with accuracies of 97%, 93%, and 94%, respectively, for the three types of data-test: original, augmented, and combined. The merging of original and augmentated images for training data has shown that the model is robust enough in producing high accuracy results.
Klasifikasi Citra Penyakit Daun Tanaman Padi Menggunakan CNN dengan Arsitektur VGG-19 Rahma Shinta; Jasril; Muhammad Irsyad; Febi Yanto; Suwanto Sanjaya
SAINS DAN INFORMATIKA : RESEARCH OF SCIENCE AND INFORMATIC Vol. 9 No. 1 (2023): Jurnal Sains dan Informatika : Research of Science and Informatic
Publisher : Lembaga Layanan Pendidikan Tinggi (LLDIKTI) Wilayah X

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Penurunan produksi padi disebabkan oleh serangan hama dan penyakit yang biasa terdapat pada bagian daun. Penelitian terkait klasifikasi jenis penyakit daun padi telah banyak dilakukan. Penelitian ini menerapkan metode Convolutional Neural Network (CNN) dengan arsitektur VGG-19 untuk klasifikasi citra penyakit daun tanaman padi. Tujuan penelitian ini adalah untuk membandingkan hasil akurasi pengujian dari model yang menggunakan augmentasi dan tanpa augmentasi data. Data pada penelitian ini terbagi atas 4 kelas, yaitu blast, brown spot, leaf smut, dan healthy dengan jumlah data asli sebanyak 440 dan data augmentasi sebanyak 1320 citra. Hasil pengujian menunjukkan bahwa akurasi tertinggi menggunakan augmentasi data yang diperoleh sebesar 94.31%, sedangkan akurasi tertinggi tanpa augmentasi data yang diperoleh sebesar 93.18%. Hasil penelitian menunjukkan bahwa augmentasi dapat meningkatkan hasil akurasi. Penggunaan optimizer Nadam menghasilkan nilai akurasi yang lebih tinggi dibandingkan Adamax. Hyper Parameter yang digunakan juga berpengaruh terhadap hasil akurasi pengujian.
Image Classification of Beef and Pork Using Convolutional Neural Network Architecture EfficienNet-B1 Isnan Mellian Ramadhan; Jasril - Jasril; Suwanto Sanjaya; Febi Yanto; Fadhilah Syafria
Indonesian Journal of Artificial Intelligence and Data Mining Vol 6, No 1 (2023): Maret 2023
Publisher : Universitas Islam Negeri Sultan Syarif Kasim Riau

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.24014/ijaidm.v6i1.21843

Abstract

The increasing demand for beef has made many meat traders mix beef with pork to get more profit. Mixing beef and pork is harmful, especially for Muslims. In this study, the EfficientNet-B1 Convolutional Neural Network (CNN) approach was used to classify beef and pork. Experiments were conducted to compare accuracy using original data (without data augmentation) and with data augmentation. The data augmentation techniques used are rotation and horizontal flip. The total dataset after the data augmentation process is 3000 images. Many different settings were tested, including learning rates (0.00001, 0.0001, 0.001, 0.01, 0.1), batch size (32, 64), and optimizer (Adam, Adamax). After testing the Confusion Matrix, the highest accuracy results were obtained using data augmentation with a batch size of 32 of 98%. Meanwhile, those without data augmentation were 96%
Klasifikasi Daging Sapi dan Daging Babi Menggunakan Convolutional Neural Network EfficientNet-B0 dengan Augmentasi Citra Hafez Almirza; Jasril; Suwanto Sanjaya; Lestari Handayani; Fadhilah Syafria
KLIK: Kajian Ilmiah Informatika dan Komputer Vol. 3 No. 6 (2023): Juni 2023
Publisher : STMIK Budi Darma

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30865/klik.v3i6.910

Abstract

The increase in counterfeit beef sales is in line with the growing demand for meat in Indonesia. Counterfeit meat, namely mixed beef and pork and pure pork sold as beef, can be distinguished using image classification. This study classifies pork, mixed, and beef using the Convolutional Neural Network (CNN) model of the EfficientNet-B0 architecture. This study uses the image augmentation method to augment the image with the aim of improving classification accuracy. The total original image is 900, while the total augmented image is 9000. The image data is divided using two data division ratios, namely 80:20 and 90:10. The highest classification accuracy results were obtained by a model using augmented images and a data division ratio of 90:10, with a combination of Adamax hyperparameter optimizer, Swish hidden activation, and a learning rate of 0.1, with an accuracy of 97.11%, precision of 97.14%, recall of 97.11%, and F1-Score of 97.11%. Meanwhile, the highest accuracy of the model using the original image is achieved by the model using a 90:10 division ratio with a combination of hyperparameter optimizer Adamax, hidden activation ReLU, and learning rate 0.01 with the results of accuracy 96.78%, precision 96.92%, recall 96.78%, and F1-Score 96.78%. The results show that the use of image augmentation methods can improve classification accuracy.
Estimasi Hasil Panen Ayam Pedaging Menggunakan Algoritma Regresi Linear Berganda Ahyani Junia Karlina; Muhammad Irsyad; Fitri Insani; Jasril; Eka Pandu Cynthia
KLIK: Kajian Ilmiah Informatika dan Komputer Vol. 3 No. 6 (2023): Juni 2023
Publisher : STMIK Budi Darma

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30865/klik.v3i6.920

Abstract

Data mining is the process of collecting and managing information that aims to extract important data from data. Currently data mining is used by companies to manage data but there are still many companies engaged in the livestock sector that have not used data mining to manage data. One of these companies is PT.PX which is a broiler company located in Riau, precisely in Sungai Pagar. The ever-increasing need for broiler chickens makes it difficult for chicken breeders to produce chicken according to market demand in each period. Unpredictable demand for broiler chickens makes breeders confused to determine how many chicks to produce. PT.PX still manages data using Microsoft Excel so the process is still very long and it is not certain to get accurate results. PT.PX also does not have a system for predicting broiler yields to find out how many chicken populations there will be in the next period. The existence of this data mining can help breeders to find out the number of populations to be produced for the next period. In predicting broiler yields, estimation methods can be used using multiple linear regression algorithms. Multiple linear regression was used to determine the relationship between feed, weight and age of chickens and chicken population. The information used in this research is information on harvested chickens obtained from 2019 to 2022. The results of multiple linear regression calculations at PT.PX obtained broiler yields of 12,217 populations
Klasifikasi Citra Daging Sapi dan Babi Menggunakan Convolutional Neural Network (CNN) dengan Arsitektur EfficientNet-B2 dan Augmentasi Data Deny Ardianto; Jasril Jasril; Suwanto Sanjaya; Lestari Handayani; Fadhilah Syafria
Jurnal Informatika Universitas Pamulang Vol 8, No 2 (2023): JURNAL INFORMATIKA UNIVERSITAS PAMULANG
Publisher : Teknik Informatika Universitas Pamulang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.32493/informatika.v8i2.30587

Abstract

Permintaan daging sapi Indonesia meningkat secara signifikan setiap tahun. Meningkatnya kebutuhan daging sapi ini sering dimanfaatkan oleh pedagang untuk mendapatkan untung lebih dengan cara mencampurkan daging sapi dan babi (oplosan). Membedakan daging sapi, babi, dan oplosan secara manual menggunakan penciuman dan penglihatan manusia sangatlah sulit. Untuk membantu membedakan daging tersebut dapat menggunakan teknologi yaitu pengolahan citra. Penelitian ini menggunakan Convolutional Neural Network (CNN) berarsitektur EfficientNet-B2 untuk pengolahan citra dan klasifikasi. Pada penelitian ini juga dilakukan proses augmentasi data citra untuk memperbanyak citra dengan tujuan meningkatkan akurasi. Jumlah citra asli daging sebanyak 900 telah mengalami peningkatan setelah dilakukan proses augmentasi, menjadi 9000 citra yang mencakup daging sapi, babi, dan oplosan. Dataset dibagi menjadi dua bagian, yaitu dataset pelatihan dan testing, dengan rasio perbandingan 80:20 dan 90:10. Dengan menggunakan dataset citra augmentasi dengan kombinasi optimizer Adamax, activation Swish, dan learning rate 0.1, penelitian ini menghasilkan akurasi klasifikasi tertinggi, yaitu 98,22% accuracy, 98,25% precision, 98,22% recall, 98,22% f1-score, dengan rasio perbandingan data 90:10.
Klasifikasi Daging Sapi dan Daging Babi Menggunakan CNN dengan Arsitektur EfficientNet-B4 dan Augmentasi Data Ahmad Paisal; Jasril Jasril; Suwanto Sanjaya; Lestari Handayani; Fadhilah Syafria
Jurnal Informatika Universitas Pamulang Vol 8, No 2 (2023): JURNAL INFORMATIKA UNIVERSITAS PAMULANG
Publisher : Teknik Informatika Universitas Pamulang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.32493/informatika.v8i2.30586

Abstract

Meningkatnya kebutuhan daging sapi, membuat harga daging sapi melonjak. Banyak pedagang melakukan kecurangan dengan melakukan oplos daging sapi dengan daging babi agar mendapatkan keuntungan yang lebih. Salah satu teknologi dalam bidang informatika dapat dimanfaatkan untuk membantu membedakan daging sapi, daging babi, dan daging oplosan. Dengan cara klasifikasi hal ini dapat dilakukan, penelitian ini menggunakan Convolutional Neural Network dengan arsitektur EfficietnNet-B4. Proses augmentasi data juga dilakukan pada penelitian ini untuk memperbanyak data citra, setelah di-augmentasi total citra menjadi 9000 dari 3 kelas. Pembagian dataset pada penelitian ini dibagi menjadi 2 yaitu 80% data latih dan 20% data uji serta 90% dan 10%. Proses pengujian dilakukan dengan memfokuskan model yang mendapatkan validation accuracy diatas 75% pada proses pelatihan. Hasil percobaan pada dataset 80:20 citra dengan augmentasi lebih unggul pada setiap model dibanding dengan citra asli. Sedangkan pada dataset 90:10 hasil percobaan dengan citra asli rata – rata lebih unggul dibanding citra dengan augmentasi.
Klasifikasi Citra Daging Sapi dan Daging Babi Menggunakan CNN Arsitektur EfficientNet-B6 dan Augmentasi Data M. Fadil Martias; Jasril Jasril; Suwanto Sanjaya; Lestari Handayani; Febi Yanto
Jurnal Sistem Komputer dan Informatika (JSON) Vol 4, No 4 (2023): Juni 2023
Publisher : STMIK Budi Darma

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30865/json.v4i4.6195

Abstract

In daily life, beef often serves as a staple food for humans. However, the high and expensive price of beef has prompted traders to adulterate it with pork for the sake of profit. Such adulteration has serious implications in the Islamic religion, where not all types of meat are considered halal (permissible for consumption), such as pork. As a result, consumers often remain unaware that the beef they purchase has been adulterated with pork. At a glance, both types of meat exhibit similar appearance and texture, making them difficult to differentiate. This research aims to classify beef and pork using a deep learning model with the Convolutional Neural Network (CNN) method, combined with data augmentation. The model used is EfficientNet-B6 with variations in the testing scenario. The variations include the ratio of training and testing data, learning rates, and optimizer for EfficientNet-B6. Data augmentation is performed using techniques such as random rotation, shifting, image scaling, vertical and horizontal flipping, and nearest pixel filling. Evaluation results using the confusion matrix show that the model with data augmentation achieves the highest accuracy for the classes of beef, pork, and adulterated samples at 92.00%, while the model without augmentation achieves an accuracy of 91.67%. However, from this experiment, the best scenario to avoid misclassifying pork and adulterated samples as beef can be obtained. This scenario involves a model with data augmentation, a 90:10 data split, SGD optimizer, and a learning rate of 0.01, which achieves the highest precision for the beef class at 96.05%. The research findings demonstrate that the use of data augmentation on images can improve the model's performance, and the model with data augmentation, a 90:10 data split, SGD optimizer, and a learning rate of 0.01 exhibits the best performance in classifying beef images.
KLASIFIKASI DAGING SAPI DAN DAGING BABI MENGGUNAKAN ARSITEKTUR EFFICIENTNET-B3 DAN AUGMENTASI DATA Maulana Junihardi; Jasril jasril; Suwanto Sanjaya; Lestari Handayani; Fadhilah Syafria
Jurnal Tekinkom (Teknik Informasi dan Komputer) Vol 6 No 1 (2023)
Publisher : Politeknik Bisnis Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.37600/tekinkom.v6i1.845

Abstract

The increasing demand for beef has made its price soar. the traders then mix beef with pork to get more profit. There is a technology in the field of informatics that can be used to differentiate beef, pork and mixed meat. This research was conducted to find out the difference between beef, pork and mixed meat. In this study, a deep learning convolutional neural network with the EfficientNet-B3 architecture is used for image identification to distinguish between beef and pork. 9000 images have been divided into three categories: mixed meat, pork and beef. This study compares the classification results using original data and data augmentation. The data augmentation models used are brightness, rotation, and horizontal and vertical inversion. Data is split 80:20 and 90:10 for training and testing respectively. The best results are achieved by using a division ratio of 90:10 on image data with augmentation which has a learning rate of 0.01 and Adamax Optimizer which has accuracy, precision and recall levels of 98.66%, 98.67% and 98.66%.