Dragon Fruit Maturity Detection and Quality Grading Dataset

(1) Everyone desires to procure top-notch, fresh fruits. In today's health-conscious society, individuals are meticulous about their dietary choices. They firmly believe that spoiled fruits can adversely affect their well-being. Consequently, the fruit market experiences a decline, leading to substantial economic implications. A key factor contributing to fruit spoilage is the manual process of gauging maturity in Bangladesh. Without being harvested at the appropriate time, fruits tend to decay over time. Accurate identification of ripe and unripe fruits is essential in determining the ideal harvest period. Dragon fruit stands as a significant, nutrient-rich crop widely cultivated across Bangladesh. It's estimated that considerable financial losses occur daily in Bangladesh due to the decay of dragon fruits. Therefore, an automated system for categorizing mature, immature, fresh, and defective dragon fruits is imperative to address this issue, benefiting fruit cultivators, vendors, and processing industries. (2) In this modern age, computer vision methods show significant promise in executing classification and detection assignments of this nature. (3) To create algorithms based on computer vision, a comprehensive dataset for Dragon Fruit is introduced, comprising both Maturity Detection and Quality Grading datasets. The Maturity Detection Dataset encompasses Immature and Mature Dragon Fruits, while the Quality Grading Dataset distinguishes Fresh from Defective Dragon Fruits. Classifications within this dataset were established in collaboration with an agricultural institute's domain expert. (4) A total of 3779 images of mature, immature, fresh, and defect dragon fruits were collected from the demonstration areas of three different locations in Bangladesh. Then from these original images, a total of 10,010 augmented images are produced by using flipping, width shifting, height shifting, brightening, rotating, shearing, and zooming techniques to increase the data number.