Abstract: Accurate classification of harvested mangoes based on outer appearance is crucial for maintaining quality, ensuring fair pricing, and reducing post-harvest losses. Manual grading is often time-consuming and error-prone. This paper presents an automated mango fruit grading system that utilizes deep learning algorithms to evaluate quality parameters such as texture, color, size, and surface defects. Techniques like Gabor filtering, Gray Level Co-occurrence Matrix (GLCM), and deep learning models including Convolutional Neural Networks (CNN) and Probabilistic Neural Networks (PNN) are used to extract and classify key features. The implementation is done using MATLAB R2023a. Experimental results show improved grading accuracy, reduced human involvement, and consistency in quality classification. This system contributes to smart agriculture by offering scalability, objectivity, and real-time usability.

Keywords: Mango Grading,Agriculture, Deep Learning, Image Processing, CNN, PNN, Gabor Filters, MATLAB

| DOI: 10.17148/IARJSET.2025.125224