Indian Jackfruit Variant Identification Using Deep Learning

Abstract

Discovering the variety of jack fruit is a pretty daunting task in middle parts Agri culture, particularly there are a lot many varieties could be possible to find since it has been cultivating on vast area of India. Conventional means of identification, although effective with human ingenuity tend to be labor intensive and time consuming. Deep learning models are now taking over the traditional way of variant classification utilizing color or shades. This progress encourages unique farming techniques to prosper, gain market dominance and also leads the way for conservation of biodiversity efforts. In this research, we present a deep learning methodology for automated and accurate identification of jackfruit variant. For this, we built an extensive dataset where total of 3602 images: Red — Pink — Baromashi each belongs to a category with their augmentations making it up-to 3600. A total of 6 state-of-the-art deep learning models were trained and evaluated on this dataset (Xception, VGG19, ResNet50, InceptionV3, MobileNetV2 & DenseNet201). The accuracy of MobileNetV2 was the highest (85.20%) followed by DenseNet201 82.42%, which are powerful methods for this application Our results demonstrate how deep learning models such as these can greatly enhance the precision and efficacy of jackfruit variant identification, providing significant implications for agricultural supply chain traceability planning as well trade & conservation. This is an important indication of future research and practical application in agriculture using the deep learning approach.