An Enhanced Grape Quality Prediction Using Near Optimal Convolution Neural Network and Image Processing Techniques

Ensuring the quality of grapes is essential for producing premium wine. Grapes that are not sweet, excessively acidic, diseased, hybrid, or unripe can adversely affect the winemaking process. This research aims to develop an advanced image processing system to identify and exclude unsuitable grapes before they are used in production. Utilizing image processing techniques and machine learning approaches, the system analyzes grape images to detect issues such as low sugar content, high acidity, diseases, hybrid traits, and ripeness levels. The model is trained on a diverse dataset of grape images captured at various stages of growth and quality. Initial results show that the model can accurately classify grapes based on these factors, ensuring only highquality grapes are selected. As a continuation of this research, the major project will advance by designing a sensor-based system for real-time grape detection and automatic picking. This integration will further optimize vineyard management and enhance winemaking efficiency. Future efforts will focus on refining the image processing model, developing the sensor system, and implementing real-time vineyard applications.