Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Varroa destructor detection on honey bees using hyperspectral imagery

Zina-Sabrina Duma, Tomas Zemcik, Simon Bilik, Tuomas Sihvonen, Peter Honec, Satu-Pia Reinikainen, Karel Horak

TL;DR

This study demonstrates that hyperspectral imagery can be used to detect Varroa destructor on honey bees by applying unsupervised K-means++ clustering on spectral reconstructions and supervised Kernel Flows - Partial Least-Squares (KF-PLS) discrimination. By centering, scaling, and projecting spectra via PCA, the method isolates bee-mite contrasts and enables clustering with as few as $4$ spectral bands via COVPROC or $12$ bands via an $R^2$-based selection, with KF-PLS achieving robust separation using a $Matern5/2$ kernel. The approach yields reliable mite detection even when mites are on top of bees, and the authors provide a public HS dataset (BeeDS_HS) to support further research and real-time hive monitoring with simplified hardware. Overall, the work establishes a practical pathway for UV-to-visible to near-infrared spectral discrimination of parasites in beekeeping contexts and highlights the potential for optimized, low-bandwidth sensing in automated hive health monitoring.

Abstract

Hyperspectral (HS) imagery in agriculture is becoming increasingly common. These images have the advantage of higher spectral resolution. Advanced spectral processing techniques are required to unlock the information potential in these HS images. The present paper introduces a method rooted in multivariate statistics designed to detect parasitic Varroa destructor mites on the body of western honey bee Apis mellifera, enabling easier and continuous monitoring of the bee hives. The methodology explores unsupervised (K-means++) and recently developed supervised (Kernel Flows - Partial Least-Squares, KF-PLS) methods for parasitic identification. Additionally, in light of the emergence of custom-band multispectral cameras, the present research outlines a strategy for identifying the specific wavelengths necessary for effective bee-mite separation, suitable for implementation in a custom-band camera. Illustrated with a real-case dataset, our findings demonstrate that as few as four spectral bands are sufficient for accurate parasite identification.

Varroa destructor detection on honey bees using hyperspectral imagery

TL;DR

This study demonstrates that hyperspectral imagery can be used to detect Varroa destructor on honey bees by applying unsupervised K-means++ clustering on spectral reconstructions and supervised Kernel Flows - Partial Least-Squares (KF-PLS) discrimination. By centering, scaling, and projecting spectra via PCA, the method isolates bee-mite contrasts and enables clustering with as few as spectral bands via COVPROC or bands via an -based selection, with KF-PLS achieving robust separation using a kernel. The approach yields reliable mite detection even when mites are on top of bees, and the authors provide a public HS dataset (BeeDS_HS) to support further research and real-time hive monitoring with simplified hardware. Overall, the work establishes a practical pathway for UV-to-visible to near-infrared spectral discrimination of parasites in beekeeping contexts and highlights the potential for optimized, low-bandwidth sensing in automated hive health monitoring.

Abstract

Hyperspectral (HS) imagery in agriculture is becoming increasingly common. These images have the advantage of higher spectral resolution. Advanced spectral processing techniques are required to unlock the information potential in these HS images. The present paper introduces a method rooted in multivariate statistics designed to detect parasitic Varroa destructor mites on the body of western honey bee Apis mellifera, enabling easier and continuous monitoring of the bee hives. The methodology explores unsupervised (K-means++) and recently developed supervised (Kernel Flows - Partial Least-Squares, KF-PLS) methods for parasitic identification. Additionally, in light of the emergence of custom-band multispectral cameras, the present research outlines a strategy for identifying the specific wavelengths necessary for effective bee-mite separation, suitable for implementation in a custom-band camera. Illustrated with a real-case dataset, our findings demonstrate that as few as four spectral bands are sufficient for accurate parasite identification.
Paper Structure (13 sections, 6 equations, 10 figures, 1 table, 2 algorithms)

This paper contains 13 sections, 6 equations, 10 figures, 1 table, 2 algorithms.

Table of Contents

  1. Introduction
  2. Materials and Methods
  3. Collected samples of bees and Varroa mites
  4. Hyperspectral imagery
  5. Mathematical methods and workflow
  6. Spectral reconstruction and clustering
  7. Wavelength selection
  8. Results and Discussion
  9. Variation profiles
  10. Full-spectral clustering
  11. Wavelength selection
  12. Discussion on future research
  13. Conclusion

Figures (10)

  • Figure 1: Samples from the HS dataset utilizied in calibration (a,b, d, e) and testing (c, f).
  • Figure 2: Utilised camera and measurement setup.
  • Figure 3: RGB visualisations of the hyperspectral images with 204 bands for the (a) model calibration data and (b) model testing data.
  • Figure 4: Cluster formation workflow.
  • Figure 5: KF-PLS workflow.
  • ...and 5 more figures