This paper explores the concept of deep neural network models and how they are evaluated with a focus on Ayurvedic purposes within the Indian context. Ayurveda is a traditional system of Indian medicine that uses plant-based treatments and cures in a much efficient and cheaper manner making accurate identification of medicinal plants extremely crucial. ResNet50, InceptionV3, YOLOv8, DenseNet, and U-Net CNN models were used to perform identification and classification using two distinct datasets: a real-world leaf image dataset that contains raw images and a segmented dataset with pre-processed leaf images. The models display advanced image processing and machine learning algorithms to achieve accurate results in terms of correct classification and identification. InceptionV3 achieved the highest accuracy at 97% which was followed by DenseNet at 96% and YOLOv8 demonstrating a strong performance in real-time object detection tasks with precision at 94% and mAP50 at 0.973. The U-Net model showed promise in segmentation tasks but faced challenges with certain leaf classes with accuracy at 94% and mAP50 at 0.963. The results translate and conclude the potential of computer vision to contribute significantly to the field of Ayurveda by providing a reliable, scalable, and efficient method for medicinal plant identification, which can be particularly beneficial in rural and remote areas where access to expert knowledge is limited.