Mujiono, Alfinas Agung (2024) Implementasi Model Hybrid CNN-SVM Klasifikasi Citra Kondisi Kesegaran Daging Ayam. Undergraduate thesis, UPN Veteran Jawa Timur.
|
Text (Cover)
19081010069-Cover.pdf Download (2MB) | Preview |
|
|
Text (Bab1)
19081010069-Bab1.pdf Download (169kB) | Preview |
|
![]() |
Text (Bab2)
19081010069-Bab2.pdf Restricted to Registered users only until 19 January 2026. Download (491kB) |
|
![]() |
Text (Bab3)
19081010069-Bab3.pdf Restricted to Registered users only until 19 January 2026. Download (573kB) |
|
![]() |
Text (Bab4)
19081010069-Bab4.pdf Restricted to Registered users only until 19 January 2026. Download (1MB) |
|
|
Text (Bab5)
19081010069-Bab5.pdf Download (15kB) | Preview |
|
|
Text (Daftar Pustaka)
19081010069-Daftarpustaka.pdf Download (142kB) | Preview |
|
![]() |
Text (Lampiran)
19081010069-Lampiran.pdf Restricted to Registered users only until 19 January 2026. Download (261kB) |
Abstract
The increasing development of internet media provides benefits which can be felt in various activity sectors, one of which is selling activities buy. Buying and selling transactions are essentially carried out by meetings between sellers and buyers have now changed through using mobile devices by using an electronic network known as e-commerce. We can find almost all categories in e-commerce, even various e-commerce have also bought and sold food ingredients easily, such as chicken meat It is recorded that approximately 10,000 products have been sold per month at one of the e-commerce, namely Shopee. With the high number of meat purchases chickens in e-commerce increase the stock provided by sellers which makes the sales results not fully sold in the hands of consumers, so sellers inevitably have to sell chicken meat that is no longer inside fresh condition to consumers. Buyers who can only view meat stocks through photos, they must have awareness of the freshness condition of the meat chicken. Along with the development of technology, intelligent image processing systems increasingly used in image classification. Image classification is wrong a field that is in demand because it can replace human visual abilities. The image classification process can be used by applying a learning algorithm machines with supervised and unsupervised types based on characteristics, shape, texture, and color of the sample image. This research uses a hybrid model CNN-SVM as a form of implementation of supervised and unsupervised supervised. Based on this explanation, the author carried out a classification of condition images freshness of chicken meat using a hybrid CNN-SVM model to determine results from the model. The output produced in this research is accuracy and performance in the classification of chicken meat freshness conditions. Results with accuracy The best results were obtained in research that used a data sharing ratio of 80:10:10 on training data, validation data, and test data through the model training process with learning rate 0.00001, which is 95% with a precision value of 95%, recall 94.8%, and f1-score 94.9%. Then the most balanced results were obtained in the research iv using a data sharing ratio of 80:10:10 and a learning rate of 0.000001 with value accuracy of 90%, precision of 90.1%, recall of 90.1%, and f1-score of 90.1%. that model then implemented using a file with h5 format which can be classify new test data. Keywords: Image Classification, Hybrid Model, CNN-SVM, Chicken Meat
Item Type: | Thesis (Undergraduate) | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Contributors: |
|
||||||||||||
Subjects: | T Technology > T Technology (General) | ||||||||||||
Divisions: | Faculty of Computer Science > Departemen of Informatics | ||||||||||||
Depositing User: | Alfinas Agung Agung | ||||||||||||
Date Deposited: | 19 Jan 2024 10:24 | ||||||||||||
Last Modified: | 19 Jan 2024 10:24 | ||||||||||||
URI: | http://repository.upnjatim.ac.id/id/eprint/20332 |
Actions (login required)
![]() |
View Item |