Implementation of Transfer Learning Using ResNet-18 for Image-Based Garbage Classification

Abstract

Effective waste management requires accurate and efficient classification systems to support recycling and environmental sustainability efforts. This study presents the implementation of a transfer learning approach using the ResNet-18 convolutional neural network for image-based garbage classification. A publicly available garbage classification dataset consisting of multiple waste categories was utilized. The dataset was divided into training and validation sets with an 80:20 ratio. Data preprocessing included image resizing, normalization, and augmentation techniques such as random horizontal flipping and rotation to improve generalization. The pretrained ResNet-18 model was adapted by freezing convolutional layers and replacing the fully connected layer according to the number of classes. The model was trained using the Adam optimizer and cross-entropy loss function. Experimental results demonstrate that the proposed approach achieved a validation accuracy of approximately 93%, indicating strong classification performance. Analysis of the confusion matrix reveals that most classes were correctly identified, with minor misclassifications occurring between visually similar categories. These findings confirm that transfer learning with ResNet-18 provides an effective and computationally efficient solution for garbage image classification tasks.