Tomato leaf diseases
Description
This dataset consists of over 2600 images of tomato leaves collected from Khagan, Charabag, located near Daffodil International University. The images include various diseases that affect tomato plants, including viral, bacterial, and fungal infections. The dataset was classified by professionals who labeled the images according to the type of disease or healthy condition observed in the leaves. After classification, the images were compressed and their size was reduced by 80% to facilitate easier handling and analysis. This dataset is intended for use in research, diagnostics, and machine learning applications focused on plant disease recognition. Selection of Disease Types: A total of 10 different diseases affecting tomato plants were identified and categorized. These diseases include both viral, bacterial, fungal, and insect-related diseases. Image Acquisition: Over 2600 images were captured from tomato plants with visible symptoms of these diseases. The leaf samples were sourced from a variety of regions to ensure diversity in the dataset. The images were taken in natural settings, ensuring the images represent realistic field conditions. Camera Setup: All images were taken using the phone’s primary camera (iPhone 11, 12MP) to ensure consistent image quality across all samples. Focus was maintained on the specific regions where symptoms of diseases appeared, and proper lighting conditions were ensured. Disease Categories: Tomato Leaf Curl Virus: 394 images Spider Mites: 307 images Leaf Mold: 66 images Leaf Miner: 519 images Late Blight: 166 images Insect Damage: 336 images Healthy Leaves: 103 images Early Blight: 204 images Cercospora Leaf Mold: 156 images Bacterial Spot: 376 images Class 16 (Uncategorized/Other): 32 images Data Labeling: The images were manually labeled by experts or through observational methods to categorize them under specific diseases or healthy leaves. Labels were applied based on clear visible symptoms such as color changes, mold growth, leaf curling, or spots that are characteristic of specific diseases. Compression and Storage: The dataset was compressed by 80% to make it manageable for use and to facilitate faster loading during processing. Images were stored in a common image format JPG for ease of access.
Files
Steps to reproduce
The dataset was gathered to identify and classify various tomato leaf diseases to facilitate research, diagnostics, and machine learning model development for plant disease recognition. Data Sources: The dataset consists of images of tomato leaves collected from different farming environments. These images were primarily captured using an iPhone 11 with a 12MP primary camera and a 26mm f/1.8 aperture lens. Images were taken under various natural light conditions to ensure a wide range of real-world scenarios. Collection Method: Selection of Disease Types: A total of 10 different diseases affecting tomato plants were identified and categorized. These diseases include both viral, bacterial, fungal, and insect-related diseases. Image Acquisition: Over 2600 images were captured from tomato plants with visible symptoms of these diseases. The leaf samples were sourced from a variety of regions to ensure diversity in the dataset. The images were taken in natural settings, ensuring the images represent realistic field conditions. Camera Setup: All images were taken using the phone’s primary camera (iPhone 11, 12MP) to ensure consistent image quality across all samples. Focus was maintained on the specific regions where symptoms of diseases appeared, and proper lighting conditions were ensured. Disease Categories: Tomato Leaf Curl Virus: 394 images Spider Mites: 307 images Leaf Mold: 66 images Leaf Miner: 519 images Late Blight: 166 images Insect Damage: 336 images Healthy Leaves: 103 images Early Blight: 204 images Cercospora Leaf Mold: 156 images Bacterial Spot: 376 images Class 16 (Uncategorized/Other): 32 images Data Labeling: The images were manually labeled by experts or through observational methods to categorize them under specific diseases or healthy leaves. Labels were applied based on clear visible symptoms such as color changes, mold growth, leaf curling, or spots that are characteristic of specific diseases. Compression and Storage: The dataset was compressed by 80% to make it manageable for use and to facilitate faster loading during processing. Images were stored in a common image format (JPEG/PNG) for ease of access. Protocols and Software: The images were pre-processed using standard image processing techniques such as resizing, normalization, and data augmentation. For any analysis and further study, tools like Python libraries (OpenCV, PIL), or specialized software for machine learning (TensorFlow, Keras) can be used for training models to classify diseases based on image features. Reproducibility: To reproduce this research, similar methodologies can be followed. Specifically, images should be captured in field settings using a phone camera or any other image-capturing device that allows high-resolution imaging. Labeling should be done based on expert knowledge or reference materials such as plant disease guides. Compression can be adjusted depending on the file size required for further analysis.