A Generative Study of Employing Convolutional Neural Networks for Plant Categorization

Plants are required to be categorized, due to their numerous advantages and significant contribution to the ecology. It has been challenging to automate the process of plant identification. This is due to the fact that machine learning does not automate feature engineering, which makes it a lengthy procedure to apply to the problem. Although deep learning offers a chance to automate feature engineering, a further difficulty is the requirement for large datasets. After some fine-tuning, we applied our convolutional neural network on a small leaf dataset. About 99.7% of the time, it was accurate. The dataset's small size resulted in a notable overfitting of the data. In this scenario, the dataset undergoes data augmentation, whereby the images were enhanced prior to being input into the model. This is not the same as the frequent practice of using dynamic augmentation.