Pota-Toma-To leaf disease images dataset
Description
This repository contains an image dataset of Potato and Tomato leaves, curated specifically for researchers developing hybrid or advanced crop disease detection systems. The dataset provides a foundation for training and evaluating computer vision models without revealing the specific proprietary architectures used in our ongoing research. Dataset Details: Target Crops: Potato and Tomato Classes Included: Early Blight, Late Blight, and Healthy leaves. Total Images: 435 images across all categories. Repository Structure: Raw_Dataset: Contains the original RGB images collected for the study. Advanced_Processed_Dataset: Contains images that have undergone advanced pre-processing techniques (including structural enhancement and noise reduction) to improve lesion visibility for machine learning applications. Database_Description: A text file detailing the exact image distribution per class.
Files
Steps to reproduce
The 'Advanced_Processed_Dataset' was generated from the 'Raw_Dataset' using the following image enhancement steps: Noise Reduction: Discrete Wavelet Transform (DWT) was applied to the raw RGB images to reduce background noise while preserving the structural edges of the leaves. Contrast Enhancement: Contrast Limited Adaptive Histogram Equalization (CLAHE) was subsequently applied to highlight and enhance the visibility of the disease lesions. Note: No code execution is required by the user. Researchers can directly download and use the provided image folders to train their own classification models.
Institutions
- VIT Bhopal UniversityMadhya Pradesh, Kothri Kalān