Deep Learning-Based Plant Classification: VGG16 and Leaf Morphology
Vishnu Kant
Abstract
Recent developments in deep learning, especially with regard to convolutional neural networks (CNNs) such as VGG16, have greatly increased automated leaf trait-based plant species classification. This work focuses on the identification, strictly from leaf photos, of four economically significant plant species: maize, cashew, cassava, and tomato. The study approach consists of careful dataset curation, thorough pre-processing to standardize image quality and VGG16 model training applied with transfer learning approaches. Accuracy, precision, recall, and F1-score evaluation measures show how well the model can correctly identify several plant species depending on their leaf form. The result depicts how efficiently it can enhance precision farming strategies and support tasks involving disease diagnosis, crop analysis, and the forecasting of yield. This study aims to provide sustainable farming strategies and preserve the overall biodiversity, therefore adding to the enhancing the field of artificial intelligence applications in viticulture, supporting food security, sustainable agriculture, climatic conditions as well as economic growth.