India boasts a rich biodiversity of plants, and for decades, it has utilized medicinal plants in Ayurveda. However, due to a lack of species knowledge, misidentification stemming from morphological similarities, high demand driven by endangered species, and other factors, medicinal plants are susceptible to adulteration and substitution. This paper aims to address the challenges of limited species knowledge and misidentification of medicinal plants by employing state-of-art computer vision techniques, specifically deep learning through Convolutional Neural Networks (CNNs), with a data-centric approach. Despite the existence of numerous computer vision applications proposed for medicinal plant identification, they often lack accurate tailoring for specific plants and struggle to precisely generalize across plants with diverse morphological features. Our study proposes an accurate computer vision application which utilizes different deep learning CNN architectures such as MobileNet, ExceptionNet, and InceptionNet for the precise identification of 10 medicinal plants with data encompassing various morphological features. Our suggested application outperforms the other models we trained in terms of accuracy, precision, recall, and F1 score with minimal inference time, owing to its emphasis on data, data augmentation, and regularization approaches like dropout.