A Deep Learning Based Plant Analysis for Ayurvedic Applications

Ayurvedic plants are rich sources of therapeutic compounds, yet accurately identifying the plants posses significant challenges. The introduction of a deep learning approach by researchers had already taken place. A deep learning approach for the automated recognition of Ayurvedic plant species. Our approach strategically combines specialized datasets to create a comprehensive collection covering one or three distinct Ayurvedic leaves. Comprising over 9000 displaying images under diverse and realistic conditions this data set Sequentially supports the testing of CNN and Mobile NetV2 architectures. Convolution neural network for spectral feature extraction and Mobile NetV2 architecture optimized through depthwise separable convolutions and transferred learning enhancing efficiency and accuracy. Comparative analysis reveals that Mobile NetV2 achieves over 81% testing accuracy by surpassing the CNN model by 13% which means CNN model achieves 68%. This underscores the value of transferable representatives in addressing data set complexities. These findings the potential of our approach to reliability distinguish visually similar ayurvedic species, a crucial capability of practical ayurvedic applications while significant progress has been made, further optimizations in architecture tuning and data search augmentation can enhance generalization performance. In somebody hopes can provide so precise and robust platform for unlocking the Ayurvedic treasures of plant species through the integration of machine learning and botanical heritage.