Angiosperm Flora of India

An assessment of the accuracy and consistency of human perception of weed cover

Publication Type:Journal Article
Year of Publication:2010
Journal:Weed Research
Date Published:2010
ISBN Number:1365-3180
Keywords:maize, monitoring weed abundance, reliability analysis, repeatability analysis, visual estimation

Andújar D, Ribeiro A, Carmona R, Fernández-Quintanilla C & Dorado J (2010). An assessment of the accuracy and consistency of human perception of weed cover. Weed Research 50, 638–647.Summary When integrated weed management approaches are used, visual estimations of weed cover are commonly employed to quantify weed infestation levels in crops. This study assessed the reliability of visual estimation of weed cover, considering various factors which may influence the accuracy and consistency of human perception. Measurements were based on visual estimates of digital images acquired in maize fields. A total of 750 images were assessed independently by four experienced observers using three cover-abundance scales, varying the assessment date or the order in which the images were assessed. A good relationship was generally found between the visual estimates of weed cover and some objective parameters (actual weed cover, weed biomass). The analyses of reliability and repeatability showed no significant differences between the estimations performed by four different observers or by the same observer at different times, independently of the scale used. Nevertheless, the comparison between visual estimations performed by the observers and actual weed cover showed that observers tended to overestimate weed cover at low weed densities and underestimate it when densities were high. Although visual assessments of weed cover were relatively accurate to quantify weed infestation level, our results showed that human perception was not reliable enough to determine actual weed cover close to the class boundaries, both in the lower range (e.g. 5%, 10%) and in the middle range (e.g. 50%) of the scale. This may be a serious limitation for the use of visual methods in threshold-based weed management programmes.

Short Title:Weed Research
Fri, 2014-01-24 22:33 -- admin
Scratchpads developed and conceived by (alphabetical): Ed Baker, Katherine Bouton Alice Heaton Dimitris Koureas, Laurence Livermore, Dave Roberts, Simon Rycroft, Ben Scott, Vince Smith