A Comparative Study of Machine Learning Models for Bee Flora Classification: Integrating Feature Extractors with Classifiers

• Leticia Torres Pereira UFC
• Daniel Santos da Silva IFCE
• Saulo Oliveira IFCE
• Renato William Rodrigues de Souza IFCE
• Wellington Franco UFC
• Jose Maria Silva Monteiro Filho UFC

Resumo

The bee flora refers to the group of plants from which bees collect floral resources, such as pollen, nectar, or both. The potential for beekeeping production in a region is influenced by its floral coverage. Therefore, understanding the foraging resources available to bees is essential for effectively managing and optimizing apiaries. We propose developing a computational model based on machine learning to classify flower images into three categories: pollen-producing, nectar-producing, or both. The objective is to provide insights about the beekeeping production potential of a specific region. The experiments utilized a dataset of 1,145 flower images from the beekeeping flora of the Sertão Central region in Ceará, Brazil. For feature extraction, ten Convolutional Neural Networks (CNNs) were employed, while classification was performed using Support Vector Machines (SVM), K-Nearest Neighbors (KNN), Multi-Layer Perceptron (MLP), and a CNN. The study aimed to identify the best combination of feature extractors and classifier methods. The results demonstrated the efficiency of the proposed approach, with the optimal combination being ResNet50 coupled with SVM, achieving a Matthews Correlation Coefficient (MCC) of 92.94%.