Detection of Freshness of the Fruits using Machine Learning Techniques

Abstract

Survival period of the fruits after harvest is relatively short. The main objective of this research is to measure the freshness of fruits by observing their CO2 release, water vapor release, and O2 absorption after harvesting for the papaya and watermelon. They were categorized into the three groups (500g-1kg, 1kg-1.5kg, 1.5kg- 2kg) and tested in 4 selected days including the harvested day, three days after harvest, a week after, and two weeks after to observe the changes in these three factors (CO2, O2, and humidity). A CO2 sensor, an O2 sensor, and a humidity sensor was set up to detect the changes. The collected data was used to train the machine learning model (Keras Sequential Model). After entering the type of the fruit, weight, the difference of oxygen, and water vapor concentration after 45 minutes, as inputs for the model, the model will predict the freshness of the fruit as a percentage. The Accuracy of the developed model was considered to be 0.989. The results of the analysis implied that the rate of O2 absorption gradually increases after harvesting and the water vapor release gradually decreases. It is suggested to use higher sensitivity sensors to obtain accurate results.