An expert system for honeybee species identification and information retrieval

Detecting honeybee species is important for ecological and agricultural research, as it helps researchers understand their behavior, population, movement pattern, and pollination habits. The paper proposes a honey bee Identification system categorizing five subspecies: Apis Cerena Indica, Apis Mellifera, Apis Florea, Apis Dorsata, and Trigona. Input images of honeybees are preprocessed to improve quality and eliminate any noise. Data augmentation methods are used to increase the dataset size, ensuring effective model training. The VGG16 architecture, known for its success in image recognition tasks, is utilized to identify important features from the dataset. Further, Rectified Linear Unit (ReLU) and Softmax layers are added, increasing the model's efficiency. The support Vector Machine model is trained to classify 5 classes of honey bees. After training the model, accurate predictions of different honeybee species with high levels of precision and recall are made. These results prove that the system effectively identifies 5 subspecies of honeybees. This system performs exceptionally well in species classification, providing advancements in ecological and agricultural studies, by implementing VGG16 and SVM.