<![CDATA[Cracks in eggshells often occur during the distribution process, both visible and invisible to the naked eye. Cracks in eggshells are a serious concern as they can lead to contamination and health risks for consumers. This study classifies cracks in chicken eggshells based on digital images using a Convolutional Neural Network (CNN)-EfficientNet. The experiment was conducted with a sample of 300 egg images in three conditions: good, cracked, and broken, with 100 images for each condition. The images were captured using a calibrated DSLR camera with a stable background. Data preprocessing included cropping, resizing, and augmentation. The data was split in an 80:20 ratio. Hyperparameters used the Adam optimizer with 50 iterations and a batch size of 32. Model performance was evaluated using loss function metrics (sparse categorical crossentropy), accuracy, and confusion matrix. Classification using EfficientNet-B0 to B3 resulted in accuracy, precision, recall, and F1-Score of 94.52%, 95.75%, 95.71%, and 95.73%; 94.05%, 94.09%, 94.05%, and 94.02%; 94.52%, 94.56%, 94.52%, and 94.54%; and 97.14%, 97.19%, 97.14%, and 97.15%, respectively. Based on the results, classification using EfficientNet shows improved performance as the model complexity increases. The findings suggest that images of eggshell cracks can be utilized for egg quality identification and can be developed for chicken egg quality classification.]]>
Copyrights © 2025
