Answers are grounded in this paper and cite their sources.
The proliferation of counterfeit sportswear makes it difficult for consumers to objectively verify product authenticity. This study proposes an intelligent computer vision–based system to identify the authenticity of sportswear by integrating object detection and deep learning classification within an Android decision support application that also considers user data security. YOLOv8 is employed to detect salient regions in images, including garment fronts and product labels, while EfficientNetB5 classifies authenticity based on extracted visual features. The dataset comprises images of genuine and counterfeit sportswear collected from multiple locations and processed through preprocessing stages: grayscaling, zero centering, histogram equalization, standardization, and Gaussian blur to improve consistency and robustness. Evaluation uses an unseen test set with Precision, Recall, and mean Average Precision as metrics. Results demonstrate that the system effectively detects relevant objects and accurately classifies authenticity, producing bounding boxes, predicted labels, and confidence scores. The mobile application operates reliably in real time, and integrated security mechanisms enhance system resilience. Nevertheless, further dataset expansion and model optimization remain necessary to improve generalization and longterm performance in diverse realworld scenarios. Future work will explore larger datasets and cross domain deployment strategies and validation.
No takes yet. Share an insight, caveat, or question.
Ningsih et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: