Black Gram Leaf Image Dataset for Disease Detection in Field Conditions
Description
In countries of Southeast Asia, black gram (Vigna mungo) is a popular lentil for its taste, nutrient components, and health benefits. It has different types of uses, which increase its economic significance in the agricultural sector of most South Asian countries. The production of this lentil is troublesome due to its proneness to different kinds of diseases, and these diseases decrease the quality of lentils significantly, which hinders the economic profit of black gram cultivation. This lentil, produced from black gram seed, is also exported to international markets. For this reason, ensuring its sound quality is a crucial task. But leaf diseases are a great threat to its quality and cultivation. To prevent and control leaf diseases, an accurate diagnosis tool is highly important for black gram cultivation, which requires advanced datasets of leaf images. This dataset is developed for advanced disease detection of black gram, which contains 587 field images of black gram. The size of all leaf images is 1024x1024 pixels, which supports almost all computer vision techniques. These images were captured from cultivation fields of black gram under the direct supervision of a plant pathology expert. Using the Make Sense (https://www.makesense.ai/) software, this dataset was developed by utilizing image annotation. A total of 4771 bounding boxes (rectangular) are present in this dataset, which were manually marked utilizing the Make Sense software. Among 4771 bounding boxes, 976 boxes were for LA (representative of healthy leaf), 2359 boxes for LB (representative of flea beetle damage), and 1436 for LC (representative of yellow mosaic disease) label. Besides disease diagnosis, analysis of leaf diseases can also be performed using this comprehensive dataset, as it contains high-quality images of black gram leaves. Most of the existing image datasets are developed for the disease classification of black gram leaves. This dataset can be used for disease detection of black gram leaf, where multiple types of diseases can be identified from a single image. This method of disease diagnosis is more efficient than the traditional image classification approach. This dataset can play a crucial role in the advanced diagnosis of black gram leaf, which is highly important for sustainable progress in black gram cultivation. Dataset Information: # Size - 70.2 MB (Compressed) # Size - 71.2 MB (Original Folder) # Number of images - 587 # Annotation Type - Rectangular bounding boxes # Total Bounding Boxes - 4771 # Labels - 3 1. LA 2. LB 3. LC # Annotations Format 1. Single CSV file 2. VOC XML format 3. YOLO format
Files
Steps to reproduce
Several articles on black gram diseases were initially studied to identify significant diseases. Then, images were collected from different cultivation fields of black gram with the help of a plant pathology expert. After that, a dataset was generated for leaf disease detection utilizing Make Sense (https://www.makesense.ai/) software. The bounding box annotation (rectangular) was used here. The whole annotation process was done manually under the direct supervision of a plant pathology expert and a data annotation expert to ensure the usability and reproducibility of the dataset.
Institutions
- Daffodil International University
- Independent University
- Hajee Mohammad Danesh Science and Technology University