Dataset of Tomato Leaves

Published: 27-05-2020| Version 1 | DOI: 10.17632/ngdgg79rzb.1
Mei-Ling Huang,
Ya-Han Chang


This database is divided into two datasets for tomato leaf images according to different image sources. The tomato leaf images of the first dataset are selected from the PlantVillage database with ten categories (nine disease categories and one health). Each image is composed of a single leaf and a single background, for a total of 14,531 images. After combining the original tomato leaf images and deleting unnecessary categories, we then adjusted the image size from 256 * 256 to 227 * 227. Afterwards, this database is divided into five subsets of 5-fold cross-validation. The detailed categories of the first dataset are: (1) Bacterial spot, (2) Early blight, (3) Healthy, (4) Late blight, (5) Leaf Mold, (6) Septoria leaf spot, (7) Target Spot, (8) Tomato mosaic virus, (9) Tomato yellow leaf curl virus, (10) Two-spotted spider mite The second dataset is images of Taiwan tomato leaves, with six categories (five disease categories and one health). It consists of a single leaf, multiple leaves, a single background and a complex background. We have 622 original images. The size of the picture is different, and we unified the image size to 227 * 227. Then we use data augmentation method to increase the number of pictures, including clockwise rotation with 90 degrees, 180 degrees, and 270 degrees; horizontal mirroring, vertical mirroring, reducing image brightness and increasing image brightness, etc. There are 4,976 images after data enhancement. The detailed categories of the second dataset are: (1) Bacterial spot, (2) Black leaf mold, (3) Gray leaf spot, (4) Healthy, (5) Late blight, (6) Powdery mildew