Recognizing Beehives’ Health Abnormalities Based on Mobile Net Deep Learning Model

Abstract

Monitoring beehive health is a major area of interest within the field of honeybee economy. Ensuring beehives are free of problems such as Varroa destructors and hive beetles, ant problems, and missing queen represents an important challenge in the honeybee industry. Therefore, it is mandatory to have untraditional ways to diagnose these types of honeybee attacks. Artificial Intelligence (AI), computer vision, and the Internet of Things (IoT) can be integrated to develop smart systems for developing warning, prediction, and recognition systems to analyze beehives' health impacts, and conditions as well as monitor bees' behaviors and the environmental conditions inside/outside beehives. In this paper, a deep learning methodology is proposed to recognize the beehives' health abnormalities, Varroa destructors, hive beetles, ant problems, and missing queens. A novel version of the MobileNet model is developed by modifying the front layers of the mobile net model for performing the features selection phase. Three optimization algorithms are utilized and tested on a benchmark dataset of beehives, Adam optimizer, Nesterov-accelerated Adam (Nadam) optimizer, and Stochastic gradient descent (SGD) for selecting the most important features to recognize the three beehive health abnormalities. The implementation and validation results proved the efficiency of the Mobile Net using Adam optimizer in classifying beehives according to the three beehive health abnormalities (Varroa destructor and hive beetles, ant problems, and missing queen) where the model achieved testing accuracy of 95% and testing loss of 35%. In addition, the validation and comparison results confirmed the superiority of Mobile Net using ADAM optimizer in recognizing beehive health abnormalities compared to four deep learning models, Shuffle Net, Resent 50, VGG-19, and Google Net.