TomatoEbola dataset

The TomatoEbola dataset consists of images of tomato leaf diseases (both infected and healthy) collected from three farms in Yobe State, Nigeria: Dikumari, Kasaisa, and Kukareta wards. The dataset includes a total of 326 images, with 152 healthy and 174 infected. Specifically, Kasaisa contributes 122 images (52 healthy and 70 infected), Dikumari contributes 101 images (50 healthy and 51 infected), and Kukareta contributes 103 images (50 healthy and 53 infected).

Each farm's images are available in two formats: raw images and YOLOv8-labeled images. Some images contain multiple leaves, so bounding boxes are provided to accurately label each leaf. For detailed information on the number of images, total annotations, average annotations per image, image size, median image ratio, and the distribution of healthy and infected annotations,  refer to the list below. Researchers can also access the Roboflow project (link) to download the data in alternative labeling formats.

Dikumari Farm  - Images: 101  - Average Annotations per Image: 2.1  - Total Annotations: 217  - Healthy Annotations: 117  - Infected Annotations: 100  - Image Size: 2.24 MP (Megapixels)  - Image Dimensions: 1345 x 1673

Kasaisa Farm  - Images: 122  - Average Annotations per Image: 2.6  - Total Annotations: 319  - Healthy Annotations: 138  - Infected Annotations: 181  - Image Size: 0.62 MP (Megapixels)  - Image Dimensions: 667 x 906

Kukareta Farm  - Images: 103  - Average Annotations per Image: 2.4  - Total Annotations: 248  - Healthy Annotations: 164  - Infected Annotations: 84  - Image Size: 0.87 MP (Megapixels)  - Image Dimensions: 929 x 943