This study tackled the critical public health threat of counterfeit drugs in Nigeria’s pharmaceutical industry. Deceptive medicines, containing incorrect or harmful ingredients, are difficult to identify and have a significant impact on low- and middle-income countries, where estimates suggest over 10% of medicines are fake. To combat this issue, a CNN Hybrid model was developed to analyze a self-captured dataset of medicine packages in the form of images and National Agency for Food and Drug Administration Control (NAFDAC) numbers. Only 10 of Nigeria’s registered pharmaceutical brands were taken into consideration due to the availability of products. Few samples of counterfeit drugs were obtained from NAFDAC. Other samples were constructed using adapted techniques from existing studies; this was achieved by modifying the original graphics slightly to create the counterfeit logos. The proposed model leveraged pre-trained deep learning architectures, ResNet-50 and VGG16 (V16RN-50), to extract features from the images that were used for classification. The extracted features were concatenated and fed into the custom trainable dense and output layer designed to identify counterfeit and real medicines. The model achieved an impressive result in multi-classification with a training accuracy of 95.83%, a validation accuracy of 94.82%, and a test accuracy of 95.1%. Counterfeit medicine detection also yielded an excellent training accuracy of 98.8%, a validation accuracy of 98.1%, and a test accuracy of 97.3%. The findings were further strengthened by high precision, recall, and F1-score metrics of 0.951, 0.951, and 0.951 for brand recognition and 0.973, 0.973, and 0.973 for detection of counterfeit medicines respectively. Additionally, the model outperformed the benchmark studies in counterfeit detection. The research demonstrated the potential of the proposed models to detect counterfeit drugs and contribute to improved public health in Nigeria .