YOLO-v8 for verticillium disease phenotyping for cotton breeding under complex field background condition

Abstract

The soil-borne disease Verticillium wilt causes significant yield losses in cotton. Developing resistant cotton varieties is a long-term solution to manage the disease. The phytopathological parameter used to assess variety resistance is the presence or absence of infection, where transverse sectioning of infected stems is discoloured/brown. The traditional method of selecting resistant varieties involves manual scoring of visual stem discolouration after cutting. However, this process is associated with high labour costs, delayed processing time and cognitive biases. Therefore, automatic detection of resistant varieties with scalable, cost-effective, and rapid phenotyping tools is needed in cotton breeding.

Code and data availability

The paper describes ~7,200 annotated cotton stem images and YOLO-v8 models, but no public deposit of the dataset, annotations, or authors' code is stated. The only URL (github.com/ultralytics/ultralytics) is a cited third-party library, not a paper-specific asset.

No evidence-backed public reproduction asset is currently recorded.