Meat Freshness Classification Using Convolutional Neural Network (CNN) Method
Keywords:
Convolutional Neural Network, Deep Learning, Image Classification, Kaggle Dataset, Meat Freshness ClassificationAbstract
This study aims to develop a digital image-based meat freshness classification model. This model uses a Convolutional Neural Network (CNN) method to distinguish between fresh and spoiled meat automatically and objectively. The CNN was built from scratch using a Sequential architecture with three convolution blocks with 32, 64, and 128 3×3 filters, without using any pretrained architecture. The dataset used is the Meat Freshness Image Dataset, which is publicly available on the Kaggle platform. After validating the image quality, the dataset used consisted of 2,266 images; 1,815 training data images and 451 validation data images. All images went through a preprocessing stage that included resizing to 128×128 pixels, converting the color space from BGR to RGB, and normalizing pixel values to the range 0–1. The model was trained for 20 epochs with a batch size of 32. The optimizer used was Adam, with a learning rate of 0.001 and a binary crossentropy loss function. Evaluation results on validation data show that the model achieved 95.34% accuracy, 95.32% precision, 97.07% recall, 96.19% F1-score, and an Area Under the Curve (AUC) of 0.9785. These results demonstrate that a simple CNN architecture, trained without transfer learning, is capable of classifying meat freshness with excellent performance. This makes it an efficient alternative for image classification systems with relatively low computational resource requirements.
