| Name | Student ID | Universitas |
|---|---|---|
| Fatimah Fatma Syifa | M008D4KX2809 | Universitas Gadjah Mada |
| Saprina Saputri | M008D4KX3152 | Universitas Gadjah Mada |
| Raditya Arviandana | M008D4KY3146 | Universitas Gadjah Mada |
The data we used consists of image datasets with 10 different objects [1]. We grouped two of these objects, Sandwich and Donut, into a new category called "Roti" The other eight objects—Baked Potato, Burger, Crispy Chicken, Fries, Hot Dog, Pizza, Taco, and Taquito—were grouped into a new category called "Bukan Roti." Here is an overview of the data distribution our team used for this project.
| No | Object Name | Number of Training Set | Number of Test Set |
|---|---|---|---|
| 1 | Sandwich | 1500 | 500 |
| 2 | Donut | 1500 | 500 |
| 3 | Baked Potato | 375 | 125 |
| 4 | Burger | 375 | 125 |
| 5 | Crispy Chicken | 375 | 125 |
| 6 | Fries | 375 | 125 |
| 7 | Hot Dog | 375 | 125 |
| 8 | Pizza | 375 | 125 |
| 9 | Taco | 375 | 125 |
| 10 | Taquito | 375 | 125 |
The model that we used are 3 different kind of model architectures with different accuracy. We used sequential, InceptionV3, and MobileNetV2. So here the comparisons:
| No | Model Name | Size (MB) | Accuracy |
|---|---|---|---|
| 1 | Sequential | 96 | 71% |
| 2 | MobileNetV2 | 126 | 76% |
| 3 | InceptionV3 | 92 | 87.17% |
We used InceptionV3 model because It has highest accuracy model.