Image-Based Recognition of Fruit Bearing Rambutan (Nephelium Lappaceum) Using Deep Learning

Rambutan is one of the world's most essential fruits because of its nutritional and medicinal benefits. However, 40 to 60 percent of the common cultivars grown from seeds are male and do not bear fruit. The worst part is that it takes four to five years from planting for this to be discovered when the plant is already in the flowering stage. This study leverages transfer learning to fine-tune three pre-trained deep learning models, notably AlexNet, VGG16, and GoogLeNet, to enable the recognition of fruit-bearing rambutan at early stages based on their leaf images. Overall, fine-tuning deep learning models to classify the sexuality of a plant using a novel dataset is feasible, evidenced by more than 95 % accuracy across all models, even to a limited number of training images, with VGG 16 exhibited the highest with 98.40 %.