Almond Variety Classification Leveraging InceptionV3: A Deep Learning Perspective

The remarkable nutritional content and adaptability of almonds, also known as Prunus dulcis, make them a highly prized and often eaten meal. The bioenergy, cosmetics, and medical sectors also make extensive use of them. To ensure quality assessment and increase market competitiveness, it is crucial to accurately identify almond cultivars such as Ak, Nurlu, Kapadokya, and Sıra. Utilizing transfer learning techniques in conjunction with an optimized InceptionV3 model, this work has increased the accuracy of the classification of these global almond varieties. Data consistency and integrity are ensured by the 2000 meticulously shot images in the dataset, which were collected under well-watched conditions in a specially designed chamber with constant illumination. 2000 meticulously shot photos that were captured in a regulated setting with constant illumination using a custom-built chamber comprise the Almond Varieties dataset. Consistency and integrity of data are ensured. With a nominal loss value of 99.14%, 98% recall, and overall precision and recall, the model has demonstrated remarkable accuracy. In particular, the effort aims to demonstrate the effectiveness of transfer learning in agricultural contexts by optimizing deep learning models for precise variety classification of almonds. The model's ability to detect even the smallest differences between almond kinds has been demonstrated by the notable improvements in classification performance that have been achieved by improving InceptionV3 on the dataset. This method can enhance market value and strategic decision-making in the almond industry, in addition to optimizing quality evaluation procedures.