Classification of fish freshness using a convolutional neural network
Abstract At work, the freshness of rainbow trout (Oncorhynchus mykiss) was analyzed, this aspect being very important to determine its quality. The objective is to propose a computational model based on a CNN to classify the freshness of the trout based on changes in the color of their eyes and gill...
Guardado en:
| Autores principales: | , |
|---|---|
| Formato: | Preprint |
| Lenguaje: | inglés |
| Publicado: |
2024
|
| Materias: | |
| Acceso en línea: | Enlace permanente (DOI o Handle) |
| Etiquetas: |
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| Sumario: | Abstract At work, the freshness of rainbow trout (Oncorhynchus mykiss) was analyzed, this aspect being very important to determine its quality. The objective is to propose a computational model based on a CNN to classify the freshness of the trout based on changes in the color of their eyes and gills. For this purpose, a dataset of images with acquired trout was created. To obtain the results, three experiments were conducted: the first with 2 classes (days 1 and 9), the second with 3 classes (days 1, 5, and 9), and the third with 5 classes (days 1, 3, 5, 7, and 9). All experiments were carried out in Google Colab. The results were validated using a confusion matrix ROC curve. The best results were obtained by the ResNeXt5032x4d model, with 2 classes achieving an accuracy of 0.9833, with 3 classes an accuracy of 0.9222, and with 5 classes an accuracy of 0.8800. |
|---|