Raisin Dataset

Overview of the Raisin Dataset

This dataset was created to classify two raisin varieties—Kecimen and Besni—grown in Turkey. It contains 900 raisin grain samples (450 from each variety) and provides a comprehensive foundation for research in computer vision and machine learning.

Dataset Features:

1. Image Processing and Morphological Analysis:
   • The images underwent preprocessing for optimal clarity and consistency.
   • Seven key morphological features were extracted using image processing techniques.
2. Statistical Data:
   • For each feature, the dataset includes minimum, mean, maximum, and standard deviation values, enabling in-depth analysis.
3. Machine Learning Models and Performance:
   • Three classification models—Logistic Regression (LR), Multilayer Perceptron (MLP), and Support Vector Machine (SVM)—were applied.
   • The highest classification accuracy of 86.44% was achieved using the SVM model.

Advantages of the Dataset

1. Agriculture-Specific Insights:
   • Facilitates advancements in precision agriculture by automating the classification of crop varieties.
   • Enables efficient quality control for raisin production industries.
2. Machine Learning Applications:
   • Supports supervised learning models for feature extraction and classification tasks.
   • Ideal for building and testing custom machine learning pipelines.
3. Educational Value:
   • Perfect for academic research, teaching datasets, and hands-on learning in computer vision and agricultural AI.
4. Real-World Applicability:
   • Applicable in industrial processes for sorting, grading, and evaluating agricultural products.
   • Provides a foundation for future studies in agricultural image processing.

Applications of the Dataset

• Classification of agricultural products using image processing techniques.
• Development of machine vision systems for automated crop evaluation.
• Statistical analysis of morphological features in biological datasets.
• Training and benchmarking machine learning algorithms.