Wuryantoro, Mahardika Virgo (2025) Perbandingan Algoritma MobileNetV2, EfficientNet Lite, dan DenseNet-169 dalam Pengenalan Jamur Beracun Berbasis Android. Undergraduate thesis, UPN Veteran Jawa Timur.
![[img]](https://repository.upnjatim.ac.id/style/images/fileicons/text.png) |
Text (Cover)
21081010077.-cover.pdf Download (4MB) |
|
Text (Bab I)
21081010077.-bab1.pdf Download (436kB) |
|
Text (Bab II)
21081010077.-bab2.pdf Restricted to Repository staff only until 22 July 2027. Download (1MB) |
|
Text (Bab III)
21081010077.-bab3.pdf Restricted to Repository staff only until 22 July 2027. Download (1MB) |
|
Text (Bab IV)
21081010077.-bab4.pdf Restricted to Repository staff only until 22 July 2027. Download (3MB) |
|
Text (Bab V)
21081010077.-bab5.pdf Download (406kB) |
|
Text (Daftar Pustaka)
21081010077.-daftarpustaka.pdf Download (319kB) |
|
Text (Lampiran)
21081010077.-lampiran.pdf Restricted to Repository staff only until 21 July 2027. Download (582kB) |
Abstract
Mushrooms are commonly found organisms and widely used as food. However, not all types of mushrooms are safe for consumption because some contain toxins that are harmful to humans. Certain poisonous mushroom species have characteristics similar to edible ones, which often leads to poisoning cases due to misidentification, especially among the general public. Image recognition technology based on Convolutional Neural Networks (CNN) can help address this issue by automating the identification process. Although several previous studies have explored the potential of CNN for mushroom recognition, most rely on high-computing-power devices or internet connectivity, making them less practical for field use. Very few studies have examined the effectiveness of CNN models on lightweight devices such as Android smartphones. To address this gap, this study aims to develop a practical Android application capable of identifying poisonous mushrooms and to compare several CNN models suitable for mobile deployment. Three CNN architectures were selected for their efficiency on mobile devices: MobileNet V2, EfficientNet Lite, and DenseNet-169. The evaluation was based on accuracy, inference time, and resource consumption efficiency. The results show that all three models achieved accuracy above 90%, with DenseNet-169 excelling in accuracy but requiring heavier computation; EfficientNet Lite being the lightest and fastest but less stable under varying image conditions; and MobileNet V2 providing a balanced performance between accuracy, efficiency, and robustness. This study aims to fill the gap in previous research by focusing on real-world implementation on mobile devices and to offer a practical solution for the public to identify poisonous mushrooms in the field.
| Item Type: | Thesis (Undergraduate) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Contributors: |
|
||||||||||||
| Subjects: | Q Science > QA Mathematics > QA76 Computer software | ||||||||||||
| Divisions: | Faculty of Computer Science > Departemen of Informatics | ||||||||||||
| Depositing User: | Mahardika Virgo Wuryantoro | ||||||||||||
| Date Deposited: | 22 Jul 2025 06:42 | ||||||||||||
| Last Modified: | 22 Jul 2025 06:42 | ||||||||||||
| URI: | https://repository.upnjatim.ac.id/id/eprint/40334 |
Actions (login required)
 |
View Item |