Dragon Fruit Stem Disease: An Annotated High-Resolution Image Dataset for Classification and Segmentation

The Dragon Fruit Stem Disease dataset is a collection of high-quality images of dragon fruit stems that are healthy and infected. The images were captured at Sierra Agro orchard in Ramgarh, Khagrachari, Bangladesh (geographical location: 22.98123°N, 91.76185°E) during October 10 to October 12, 2024. The dataset consists of a total of 724 images, which are categorically separated into six classes: Anthracnose, Brown Stem Spot, Gray Blight, Soft Rot, Stem Canker, and Healthy. Photographs were taken using the high-resolution camera of a OnePlus 7T in natural light, with ambient temperatures ranging from 23°C to 25°C. The original images, which were recorded in JPG format at 3000x4000 pixels, have been reduced to 640x480 pixels at 72 dpi for better practicality. The dataset is also provided with meticulous annotations that provide far-reaching categorizations of the diseases seen and allow semantic segmentation analyses. This resource is intended to aid research and development in image segmentation and plant disease detection techniques.