AgriLeafNet: Fruit Tree Leaf Dataset for Agricultural Research
Description
AgriLeafNet is a labeled dataset of 24,888 annotated (1,504 original and 23384 augmented) high-resolution RGB fruit tree leaves images spread across 12 different species gathered in a rural village in Bangladesh. All the photographs were taken with a Samsung A22 5G phone with a standardized white A4 paper background and photographed to capture the two sides of every leaf so that representation is as exhaustive as possible. The collection was made into folders specific to each class and all the images are labeled carrying the corresponding scientific name. They are Syzygium cumini (Jamun), Litchi chinensis (Litchi), Mangifera indica (Mango), Ziziphus zizyphus (Jujube), Psidium guajava (Guava), Elaeocarpus serratus (Olive), Baccaurea motleyana (Burmese Grape), Phyllanthus acidus (Otaheite Gooseberry), Citrus limon (Lemon), Citrus maxima (Pomelo), Annona muricata (Soursop), and The selected classes are based on the agricultural importance and morphological peculiarities. This data can be used to provide a strong benchmark of machine learning models in agriculture training and validation. It helps researchers to generate precision farming, biodiversity tracking, and auto classification of leaves tools and it provides cost-effective, infinite, and smartphone-based frames of data collection. AgriLeafNet should be used in classification of fruits leaves, plant disease identification, environmental study, and production of artificial intelligence-based agricultural innovations, which makes the tool a useful resource to researchers and practitioners in the science of plant, ecology and agricultural engineering.