Published Paper Details:

In recent years, the increase in counterfeit products has caused serious problems for consumers and businesses, and has led to the loss of trust in e-commerce and physical stores. This study presents a new method for artifact detection using convolutional neural networks (CNN), a deep learning technique suitable for image recognition. We have created a comprehensive database containing images of structures and artifacts from multiple categories, which provides greater representation of features. We increase the robustness of the data and reduce the effects of overfitting by using advanced data enhancement techniques. Our CNN models are carefully designed to include multiple convolutional processes, followed by joint and fully connected processes, and optimized using methods such as version and batch normalization. The performance of the model was evaluated using metrics such as accuracy, precision, recall, and F1 score, resulting in significant improvements in the detection process. Our results show that the CNN-based method can identify counterfeit products with high accuracy and provides a reliable tool for consumers and retailers. This research not only contributes to the field of product recognition, but also lays the foundation for the future development of the use of deep learning in counterfeit product protection.