Automated Pharmaceutical Drug and Vitamin Classification Using the EfficientNetB3 Model in Deep Learning

In this research, the fine-tuned EfficientNetB3 model is utilised for detecting the pharmaceutical drugs and vitamins. Emphasizing the need of correct medication identification in healthcare, this paper seeks to improve patient safety and avoid drug distribution mistakes. Collected via the open-source Kaggle platform, ten different products as Alaxon, Bactidol, Bioflu, Biogesic, and others have high- resolution images in the dataset, painstakingly annotated to guarantee correct classification. Refined using batch normalization, dropout, and dense layers, the EfficientNetB3 model attained an overall accuracy of 99%, proving its ability in differentiating between like-looking pharmaceutical goods. Classification reports, confusion matrices, and training- validation loss curves among other detailed analyses show the model's strong performance over all classes. With a consistent method for automated medicine and vitamin identification, the results of this study could have major consequences for pharmaceutical management. Expanding the dataset and improving model generalization will be main priorities of next development.