Comparative Analysis of Classical Machine Learning and Deep Learning Methods for Fruit Image Recognition and Classification

In this investigation, the crucial role of fruits in daily lives is acknowledged, with emphasis placed on their significance in nutrition and agriculture. The primary focus is directed towards fruit image recognition and classification, a task of paramount importance in the present context. To expound on the methodology, classical machine learning approaches, encompassing K -Nearest Neighbors (KNN), Support Vector Machines (SVM), and Decision Trees (DT), are leveraged. Additionally, the capabilities of deep learning are harnessed through the utilization of the AlexNet model. The dataset selected, Fruit -360, is widely acknowledged and utilized, underscoring its popularity and relevance within the research community. Of particular note in the findings is the exceptional performance of the AlexNet model, with the highest metrics in accuracy (99.85%), precision (99.92%), sensitivity (99.86%), and an impressive F1 score (99.89%) when compared to all tested algorithms. The effectiveness of deep learning, especially in tasks revolving around imagebased classification, is underscored by these results. The impact of these noteworthy results transcends multiple domains. In agriculture, the potential for automated fruit sorting holds the promise of heightened efficiency and decreased waste. Similarly, in healthcare, the integration of fruit recognition into dietary and nutritional assessments presents a substantial opportunity. A thorough outlook on the progression of fruit recognition and classification is encapsulated by this study, offering a positive outlook for the future in these fields.