Claim Missing Document
Check
Articles

Found 3 Documents
Search
Journal : Journal of Computer Science, Information Technology and Telecommunication Engineering (JCoSITTE)

Analysis of the Naïve Bayes Method in Classifying Formalized Fish Images Using GLCM Feature Extraction Ayu Pariyandani; Eka Pirdia Wanti; Muhathir Muhathir
Journal of Computer Science, Information Technology and Telecommunication Engineering Vol 1, No 2 (2020)
Publisher : Universitas Muhammadiyah Sumatera Utara, Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (655.435 KB) | DOI: 10.30596/jcositte.v1i2.5171

Abstract

Fish is one of the foods that are high in protein so that many Indonesians consume fish as protein intake for health. Fish can be found in any waters including Indonesian marine waters, so that some of the Indonesian people work as fishermen. This causes the number of fish catches to increase and the fishermen have to sell the fish quickly in at least one day because the fish will rot easily if not consumed immediately. This has led some traders to cheat by mixing formaldehyde with fish that are not sold out. This action is very detrimental to consumers, so they must be more vigilant in choosing or buying fish on the market. One way for consumers to recognize formaldehyde fish is a technology that can distinguish fresh fish or formalin fish based on the image of the fish, Naive Bayes and GLCM (Gray Level Co-Occurrence Matrix) by using this method the accuracy of this system can reach up to 70%.
Performance Comparison of Boosting Algorithms in Spices Classification Using Histogram of Oriented Gradient Feature Extraction Muhathir Muhathir; Reydo Trisno Pangestu; Ira Safira; Melisah Melisah
Journal of Computer Science, Information Technology and Telecommunication Engineering Vol 4, No 1 (2023)
Publisher : Universitas Muhammadiyah Sumatera Utara, Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30596/jcositte.v4i1.13710

Abstract

Spice classification is an important task in the food industry to ensure food safety and quality. This study focuses on the classification of spices using the HoG feature extraction method and boosting algorithms. The objective of this research is to compare the performance of four different models of boosting algorithms, namely Adaboost Classifier, Gradient Boosting Classifier, XGB Classifier, and Light GBM Classifier, in classifying spices. The evaluation metrics used in this research are Precision, Recall, F1-Score, F2-Score, Jaccard Score, and Accuracy. The results show that the XGB Classifier model achieved the best performance, with a precision of 0.811, recall of 0.809, and F1-score of 0.809, while the Adaboost Classifier model had the lowest performance, with a precision of 0.709, recall of 0.689, and F1-score of 0.682. Overall, the results indicate a fairly good success rate in classifying spices using the HoG feature extraction method and boosting algorithms. However, further evaluation is needed to improve the accuracy of the classification results, such as increasing the number of training data or considering the use of other feature extraction methods
Hyperparameter Model Architecture Xception in Classifying Zophobas Morio and Tenebrio Molitor Amri Ismail Tumanggor; Muhathir Muhathir
Journal of Computer Science, Information Technology and Telecommunication Engineering Vol 4, No 2 (2023)
Publisher : Universitas Muhammadiyah Sumatera Utara, Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30596/jcositte.v4i2.15800

Abstract

Zophobas Morio and Tenebrio Molitor are popular larvae as feed ingredients that are widely used by animal lovers to feed reptiles, birds, and other poultry. However, these two larvae are similar in appearance; their nutritional content is very different. Zophobas Morio is more nutritious and has a higher economic value compared to Tenebrio Molitor. Due to limited knowledge, many animal lovers have difficulty distinguishing between the two. This study aims to build the best configuration of the Xception architecture hyperparameter model that can distinguish between the two. The model is trained using images taken from mobile phones. Training is carried out using the parameters Epoch 15, Batch Size 32, Optimizer Adam, RMSprop, and SGD. The experimental results on the dataset show that the best accuracy for the Xception architecture hyperparameter model is Optimizer Adam with an accuracy rate of 100%, and Optimizer SGD with an accuracy rate of 100%. And of course, it gives very good results