Study and implementation of ML model for fruit quality classification in agriculture domain

Abstract

Artificial Intelligence, Machine Learning, and Deep Learning have restructured the modern years of enhancement by fabricating momentous influence throughout the society. The advancements of technology in the field of ML and DL made Artificial Intelligence attractive and applicable to various areas. The applications of ML and DL algorithms are numerous. The ML and DL algorithms are used to solve complex problems in various domains like cyber security, healthcare, agriculture, banking application etc.. Agriculture is one of the most important sectors in India as far as the GDP or jobs created sectors are concerned. More than 70 percentage people in India are directly or indirectly depended on the agriculture. Fruit market contributes large share in the profit of agriculture. India is the second largest fruit producing country in the world. It is also ranked at top in the fruit exporter list. Fast and accurate fruit classification is the need of fruit market and stakeholders like farmers, fruit industries, retailer and customers. Building a machine learning model for fast and accurate classification of fruits with quality parameter has emerged as an important research topic. newlineAll the activities in the agriculture domain are broadly categorized into pre-harvesting, harvesting, and post-harvesting activates. This research work presented the in-depth and systematic survey of applications of machine learning algorithms in each phase of agriculture. Important parameters which are considered while building the ML models by other researchers are investigated and listed. We presented the current challenges, gaps and solution to reduce them while building machine learning models for fruit classifications. Misclassification is the major problem which is overlooked by many researchers. We proposed a unique framework called MNet: Merged net which not only address the misclassification but also improve the accuracy and make ML model modular.