Abdurrahman, Nizar (2024) PERBANDINGAN PERFORMA KLASIFIKASI CITRA IKAN MENGGUNAKAN METODE K-NEAREST NEIGHBOR (KNN) DAN CONVOLUTIONAL NEURAL NETWORK (CNN). Undergraduate thesis, UPN VETERAN JAWA TIMUR.
|
Text (COVER)
19081010025-COVER.pdf Download (1MB) | Preview |
|
|
Text (BAB1)
19081010025-BAB1.pdf Download (338kB) | Preview |
|
![[img]](http://repository.upnjatim.ac.id/style/images/fileicons/text.png) |
Text (BAB2)
19081010025-BAB2.pdf Restricted to Registered users only until 19 January 2026. Download (494kB) |
|
|
Text (BAB3)
19081010025-BAB3.pdf Restricted to Registered users only until 19 January 2026. Download (701kB) |
|
|
Text (BAB4)
19081010025-BAB4.pdf Restricted to Registered users only until 19 January 2026. Download (1MB) |
|
|
Text (BAB5)
19081010025-BAB5.pdf Download (323kB) | Preview |
|
|
Text (DAFTARPUSTAKA)
19081010025-DAFTARPUSTAKA.pdf Download (320kB) | Preview |
Abstract
Fish, as cold-blooded animals with a characteristic spine, gills, and fins, are very dependent on water as a medium in which they live. Difficulty in determining the type of fish, especially by young people, such as difficulty distinguish catfish, catfish, tilapia, and gourami which have similar shapes, to be basic research using digital image technology. Classification, as a process grouping based on characteristics, becomes the main focus in context digital image. The research compares the performance of fish image classification between the K NN and CNN methods. Previously, K-NN was better than Naïve Bayes in image classification. This research retests the K-NN method and compare it with CNN to determine which method is superior in classification. Fish image data of various types will be classified using both methods, and the results will be compared using Confusion Matrix. Performance of k-Nearest Neighbor (k-NN) and Convolutional Neural methods Network (CNN) in fish image classification compared to fish datasets. With RGB feature extraction, both tested five times with varying ratios of training data and data test. The results show that CNN has higher accuracy, especially on division of training data 80% and test data 20%, with accuracy reaching 88%. In contrast, k-NN achieves the highest accuracy at a training data split of 90% and 10% test data, with 72% accuracy. Thus, it is concluded that CNN is more superior in fish image classification, especially in the ratio of training data and test data 80%:20%. Keywords: CONVOLUTION NEURAL NETWORK (CNN), Fish, Classification, K NEAREST NEIGHBOR (K-NN), Performance Comparison
| Item Type: | Thesis (Undergraduate) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Contributors: |
|
||||||||||||
| Subjects: | T Technology > T Technology (General) | ||||||||||||
| Divisions: | Faculty of Computer Science > Departemen of Informatics | ||||||||||||
| Depositing User: | Nizar Abdurrahman | ||||||||||||
| Date Deposited: | 19 Jan 2024 08:41 | ||||||||||||
| Last Modified: | 19 Jan 2024 08:41 | ||||||||||||
| URI: | http://repository.upnjatim.ac.id/id/eprint/20254 |
Actions (login required)
 |
View Item |