Date fruit detection dataset for automatic harvesting

Abstract

Date fruit is one of the most beneficial and delicious fruits in the world through its nutritional value, it's a rich source of sugar, Protein, Manganese, Fibers, and many other vitamins, on the other hand, date production in the world reached up to 9.45 million tons, one palm tree can produce dates for 40–50 years which makes date production one of the main pillars of the economy. Generally, the methods used for pre and post-harvesting are done manually, making the operation heavy in terms of time and economy. Research in the area of harvesting automation is still limited due to the absence of data. This work presents a full dataset for detection, classification, analysis, and harvesting decisions. The dataset contains a collection of images and videos for different varieties of dates at different stages of maturity, taking into consideration all the natural conditions including light, contrast, dates in bags, and multi-scaled images. The big advantage of this dataset is it can help to put agriculture research at an advanced level by robotizing all the pre and post-harvesting tasks. The dataset is published for free in ZENODO [1] https://zenodo.org/doi/10.5281/zenodo.8315234.

Keywords

• Date fruits
• Intelligent harvesting
• Fruit classification
• Fruit detection