Hyperspectral imaging for pest symptom detection in bell pepper

Abstract

Background: The automation of pest monitoring is highly important for enhancing integrated pest management in practice. In this context, advanced technologies are becoming increasingly explored. Hyperspectral imaging (HSI) is a technique that has been used frequently in recent years in the context of natural science, and the successful detection of several fungal diseases and some pests has been reported. Various automated measures and image analysis methods offer great potential for enhancing monitoring in practice.

Results: In this study, the use of hyperspectral imaging over a wide spectrum from 400 to 2500 nm is investigated for noninvasive identification and the distinction of healthy plants and plants infested with Myzus persicae (Sulzer) and Frankliniella occidentalis (Pergande) on bell peppers. Pest infestations were carried out in netted areas, and images of single plants and dissected leaves were used to train the decision algorithm. Additionally, a specially modified spraying robot was converted into an autonomous platform used to carry the hyperspectral imaging system to take images under greenhouse conditions. The algorithm was developed via the XGBoost framework with gradient-boosted trees. Signals from specific wavelengths were found to be associated with the damage patterns of different insects. Under confined conditions, M. persicae and F. occidentalis infestations were distinguished from each other and from the uninfested control for single leaves. Differentiation was still possible when small whole plants were used. However, application under greenhouse conditions did not result in a good fit compared to the results of manual monitoring.

Conclusion: Hyperspectral images can be used to distinguish sucking pests on bell peppers on the basis of single leaves and intact potted bell pepper plants under controlled conditions. Wavelength reduction methods offer options for multispectral camera usage in high-grown vegetable greenhouses. The application of automated platforms similar to the one tested in this study could be possible, but for successful pest detection under greenhouse conditions, algorithms should be further developed fully considering real-world conditions.

Keywords: Automated; Monitoring; Noninvasive; Pest detection; Robot.

Fig. 1

Greenhouse test setup. Test setup with three rows in a soil greenhouse in 2022. The thin lines indicate the rails on which the mobile HSI measurement platform was driven. The marked rows of symbols represent the individual plants, and the symbol shapes represent the respective plant treatments

Fig. 2

HSI measurement platform. Self-constructed mobile HSI measurement platform on rails in the experimental greenhouse with bell peppers grown in double rows. The stand consists of 300 W halogen spotlights (1–3), two hyperspectral cameras, a Hyspex Vis-NIR at 400,950 nm (4), a Hyspex NIR-SWIR at 950 bis 2500 nm (5), the control computer (6), the input device (7), the spraying robot control (8) and an angle mirror (9)

Fig. 3

Visualization and segmentation of images. Example of a 3-channel visualization of an SWIR image (left) and a VNIR image (middle), together with the resulting leaf segmentation mask created via deep learning (right)

Fig. 4

Visualization of tiles. Visualization of the different mean values for a single wavelength in a tiled image

Fig. 5

Influence of wavelength number on accuracy. Changes in the accuracy of the three-class prediction algorithm under controlled conditions when reducing the number of wavelength channels used. Analyses were carried out via the XGBoost integrated functions with images of the top and bottom sides of leaves

Fig. 6

Comparison of visual and algorithm results under greenhouse conditions. Visual estimates based on insect count categories from 0 (no insect) to 9 (> 100 insects) (gray) and algorithm estimates given as the percentage tiles with > 80% predicted infestation probability (black). The different treatments are shown separately, as are the infestations of M. persicae and F. occidentalis in each treatment