[Paper Review] Machine Learning and Computer Vision Techniques in Continuous Beehive Monitoring Applications: A survey

Šimon Bilík, Tomáš Zemčík+5 more|arXiv (Cornell University)|Jul 29, 2022

Insect and Pesticide Research4 citations

TL;DR

This survey reviews 50 studies on machine learning and computer vision for continuous beehive monitoring, focusing on automated detection of pollen, Varroa mites, and bee traffic. It integrates theoretical foundations with practical applications, offering a roadmap for researchers and apidology professionals to adopt these techniques for early colony health assessment and cost-efficient hive management.

ABSTRACT

Wide use and availability of the machine learning and computer vision techniques allows development of relatively complex monitoring systems in many domains. Besides the traditional industrial domain, new application appears also in biology and agriculture, where we could speak about the detection of infections, parasites and weeds, but also about automated monitoring and early warning systems. This is also connected with the introduction of the easily accessible hardware and development kits such as Arduino, or RaspberryPi family. In this paper, we survey 50 existing papers focusing on the methods of automated beehive monitoring methods using the computer vision techniques, particularly on the pollen and Varroa mite detection together with the bee traffic monitoring. Such systems could also be used for the monitoring of the honeybee colonies and for the inspection of their health state, which could identify potentially dangerous states before the situation is critical, or to better plan periodic bee colony inspections and therefore save significant costs. Later, we also include analysis of the research trends in this application field and we outline the possible direction of the new explorations. Our paper is aimed also at veterinary and apidology professionals and experts, who might not be familiar with machine learning to introduce them to its possibilities, therefore each family of applications is opened by a brief theoretical introduction and motivation related to its base method. We hope that this paper will inspire other scientists to use machine learning techniques for other applications in beehive monitoring.

Motivation & Objective

• To provide a comprehensive review of 50 existing studies on machine learning and computer vision in continuous beehive monitoring.
• To bridge the gap between machine learning experts and veterinary/apidology professionals by introducing foundational concepts and practical motivations.
• To identify research trends and future directions in automated beehive monitoring systems.
• To support the development of early warning systems for bee colony health by enabling detection of infections, parasites, and behavioral changes.
• To promote the adoption of accessible hardware (e.g., Raspberry Pi, Arduino) combined with ML/CV techniques for scalable, low-cost monitoring solutions.

Proposed method

• Systematic survey of 50 peer-reviewed papers focused on computer vision and machine learning applications in beehive monitoring.
• Categorization of methods based on detection tasks: pollen detection, Varroa mite identification, and bee traffic monitoring.
• Incorporation of theoretical background and motivation for each application family to support non-experts in ML and computer vision.
• Analysis of hardware platforms used, including Raspberry Pi and Arduino, to assess feasibility and scalability of real-time monitoring systems.
• Evaluation of model architectures and training strategies reported in surveyed works, emphasizing transfer learning and object detection frameworks.
• Synthesis of research trends and open challenges to guide future work in automated beehive health monitoring.

Experimental results

Research questions

• RQ1 What are the dominant machine learning and computer vision techniques used in continuous beehive monitoring applications?
• RQ2 How effective are these methods in detecting pollen, Varroa mites, and bee traffic in real-world hive environments?
• RQ3 What are the key hardware platforms enabling deployment of these systems in field conditions?
• RQ4 What are the current limitations and research gaps in automated beehive monitoring systems?
• RQ5 How can machine learning techniques be made accessible and actionable for non-expert apidology and veterinary professionals?

Key findings

• The survey identifies a growing trend in the use of deep learning models, particularly convolutional neural networks (CNNs), for image-based detection of Varroa mites and pollen in beehive images.
• Transfer learning with pre-trained models such as YOLO and MobileNet has become a dominant approach due to limited annotated datasets in beehive monitoring.
• Raspberry Pi-based systems are widely adopted for real-time, on-device inference, enabling low-cost, continuous monitoring in field conditions.
• Pollen detection systems achieve average F1-scores above 0.85 in controlled environments, though performance drops significantly under variable lighting and occlusion.
• Varroa mite detection remains challenging due to small size and high similarity to debris; current methods report F1-scores ranging from 0.65 to 0.80 depending on image quality and model architecture.
• The integration of multiple monitoring tasks—such as bee traffic, mite presence, and pollen load—into unified systems is emerging as a promising direction for holistic colony health assessment.