The Asian subcontinent produces a vast range of fruits throughout the seasons. However, correctly classifying these fruits according to their qualities can be difficult, frequently necessitating the knowledge of fruit experts and cutting-edge equipment to produce accurate results. Therefore, to enable sophisticated grading methods that efficiently sort and evaluate fruit quality based on various characteristics (such as form, color, size, texture, and other crucial parameters), A unique dataset is deployed to support advanced grading systems. This dataset helps researchers explore genetic variation, ecological adaptation, and environmental factors that affect fruit qualities for conservation and sustainable agriculture. Using a mobile camera, these data are personally collected at various times of the day at local markets in Bangladesh that receive optimal sunlight. To create a unique dataset, 6 types of fruit consisting of 3,167 images have been collected. These six different types of fruit: apple, banana, burmese grape, mango, papaya, and tomato were used for quality grading, categorizing them as (i) good, (ii) medium, and (iii) bad. This dataset will help researchers in biodiversity conservation by building efficient machine-learning models and applying machine-learning techniques. Smart fruit grading, classification, and yield prediction automation systems can be built with this dataset.