Skip to main content

Table 2 Accuracy assessment of pixel-based classification method for plant and soil classification

From: Soybean iron deficiency chlorosis high-throughput phenotyping using an unmanned aircraft system

 

Reference data

User accuracy (%)

Soil

Plant

Row total

Predicted data

 Soil

469

31

500

93.8

 Plant

13

487

500

97.4

 Column total

482

518

1000

 

 Producer accuracy (%)

97.3

94.0

  

Overall accuracy (%) = 95.6

  1. One thousand random points were generated and placed on the orthomosaiced image using the equalized random sampling method. Predicted data were generated using k-means clustering and reference data were manually created using visual observations of the images. Accuracy assessment results were generated using ERDAS IMAGINE software. An overall classification accuracy of 95.6% was achieved
  2. The diagonal elements are italicized to highlight the number of correctly classified pixels in terms of plant or soil classifications
Back to article page

Plant Methods

ISSN: 1746-4811