Machine Learning for Detection and Severity Estimation of Sweetpotato Weevil Damage in Field and Lab Conditions

TL;DR

A computer vision-based approach for the automated evaluation of weevil damage in both field and laboratory contexts is introduced and indicates that computer vision technologies can provide efficient, objective, and scalable assessment tools that align seamlessly with contemporary breeding workflows.

Abstract

Sweetpotato weevils (Cylas spp.) are considered among the most destructive pests impacting sweetpotato production, particularly in sub-Saharan Africa. Traditional methods for assessing weevil damage, predominantly relying on manual scoring, are labour-intensive, subjective, and often yield inconsistent results. These challenges significantly hinder breeding programs aimed at developing resilient sweetpotato varieties. This study introduces a computer vision-based approach for the automated evaluation of weevil damage in both field and laboratory contexts. In the field settings, we collected data to train classification models to predict root-damage severity levels, achieving a test accuracy of 71.43%. Additionally, we established a laboratory dataset and designed an object detection pipeline employing YOLO12, a leading real-time detection model. This methodology incorporated a two-stage laboratory pipeline that combined root segmentation with a tiling strategy to improve the detectability of small objects. The resulting model demonstrated a mean average precision of 77.7% in identifying minute weevil feeding holes. Our findings indicate that computer vision technologies can provide efficient, objective, and scalable assessment tools that align seamlessly with contemporary breeding workflows. These advancements represent a significant improvement in enhancing phenotyping efficiency within sweetpotato breeding programs and play a crucial role in mitigating the detrimental effects of weevils on food security.

Figures (9)

• Figure 1: Example field plots illustrating the adapted 5-point weevil damage severity scale. Severity scores (1, 3, 5, 7, 9) are assigned based on the proportion of damaged roots relative to the total number of roots in each plot.
• Figure 2: Orthogonal views of storage roots captured at the plot level. The left image shows the top view, taken as the roots were initially arranged. The right image shows the bottom view, captured after rotating the roots to reveal the underside.
• Figure 3: Tin containers used for the controlled infestation process. Each container held a single root and 5–10 adult weevils, sealed and incubated for 24 hours.
• Figure 4: Sweetpotato genotype ‘Tanzania’ showing two zoomed-in views. The green boxes (top right) highlight feeding holes with fecal matter, identifiable by whitish, sponge-like tissue. These white tissues contain weevil eggs. The blue boxes (bottom right) show feeding holes without fecal matter, which appear as clean surface depressions. Accurately detecting and estimating the number of feeding holes provides an indicator of a genotype's resistance — fewer holes suggest higher resistance to sweetpotato weevil damage.
• Figure 5: Three viewable sections (A, B, and C) of a single sweetpotato root after colour-coded segmentation. This approach enabled systematic imaging to avoid duplication and ensure full coverage of the sweetpotato root surface for accurate damage assessment.