GRAPE QUALITY PREDICTION WITH PRE-POST HARVESTING USING FUSION DEEP LEARNING

Abstract

In modern society, the agriculture industry is crucial. Agriculture industry has to support the load of increasing population, year by year. Implementing robotics in agriculture can heavily rely on computer vision and perception techniques. Machine learning is very useful for object recognition and picture categorization. In this work, machine learning was used to identify grape bunches in vineyards at various phases of development, including the early stage right after bloom and the middle stage, when the grape bunches display an intermediate degree of development. The dataset is suggested in this investigation because the training inputs are not freely available. So we made a dataset. Separate metrics used to benchmark and explain the models. The results revealed that the built models successfully detect grape clusters from pictures. The suggested system's performance was adequate when considering the approach's minimal resource utilisation, cheap cost, and minimum power hardware device requirements, which allow for simpler models