Combining computer vision and deep learning to classify varieties of Prunus dulcis for the nursery plant industry

Varietal control to avoid unwanted varietal mixtures is an important objective for the nursery plant industry. In this study, we have developed and analyzed capabilities of a computer vision system based on deep learning varieties in industry evaluating its capabilities. For purpose, three datasets images were compared. The came from two almond trees (Prunus dulcis) named Soleta Pentacebas. Each dataset contained with different scales: whole plant, leaf, venation. Gradient-weighted Class Activation Mapping (Grad-CAM) technique was used unveil most features discriminate between both varieties. provided classification accuracies above 97% test set, being leaf dataset, 98.8% accuracy, one providing best results. Concerning plants, Grad-CAM showed that they are located center is, conclusion, shown promising mixtures.