Comparative Analysis of ResNet50, and VGG16 Architectures for Counterfeit Logo Identification

The unregulated proliferation of counterfeit branding in the era of digital technology poses a significant risk to the worth of brands and erodes the confidence of consumers. In order to tackle this particularular difficulty, the present study focused on the domain of counterfeit logo identification, with a specific emphasis on three widely recognized International brands: Adidas, Nike, and The North Face. In this study, a comprehensive comparative analysis was performed on the VGG16 and ResNet50 deep learning architectures using advanced machine learning techniques to identify counterfeit logos. The models underwent training and validation using a carefully selected dataset that included authentic and fraudulent variations of logos from the brands. The empirical findings underscored the superior performance of the VGG16 model, which achieved an accuracy rate of 77%, outperform the 72% accuracy rate achieved by the ResNet50 model. Moreover, the VGG16 model regularly demonstrated superior performance in evaluation criteria such as precision, recall, and F1-score. The findings highlight the complexities and difficulties associated with detecting counterfeit logos, while also demonstrating the promise of artificial intelligence in protecting the integrity of brands. The findings of this study have important implications for brand protection and customer trust in the digital marketplace, considering the widespread familiarity with the companies under investigation.