Interpreting Data from your PlantCV pipeline

Introduction

Before results can be interpreted, you must first convert your output JSON file into CSV tables. See Data Output: Converting output JSON format to CSV tables.

Interpreting data from your PlantCV pipeline

After you have run an image analysis pipeline and converted the JSON file output to CSV tables, you should see one or more of the following .csv files in your working directory folder:

  • Single-value traits file: “results-table-single-value-traits.csv

  • Multi-value traits file: “results-table-multi-value-traits.csv

The files that you see is dependent on the type of data that you extracted from your pipeline.

Single-value traits

In the single-value traits file, the data will be organized by image file names. These file names are parsed into separate columns (camera, timestamp, ID, etc.) as specified in your configuration file (as discussed in Running Pipelines: Using your PlantCV Pipeline to Batch Process Images).

The single-value traits file can include data for the following types of measurements:

Geometric measurements

Geometric measurements for individual objects (such as seeds) are identified in each image according to area, perimeter, width, height, etc. (Table 1).

Table 1.

Scale-marker measurements

Scale-marker measurements are provided for each image (Table 2).

Table 2.

Measurements you may have created

If you created measurements using the pcv.outputs.add_observation() command in your pipeline (as shown in code below from the example pipeline in Figure 2), these will also be included in the single-value traits file.

 

Multi-value traits

The multi-value traits file (Table 4) contains color data, which may be given on a basis of the whole image or for individual objects within the image.

RGB color analysis

RGB color data are presented in the multi-value traits file as the relative frequency (percentage) of pixels in each channel for gray values from 0 to 255.

L*a*b* color analysis

L*a*b* color data are presented in the multi-value traits file as 0 to 100 for lightness (L*), and as -128 to 127 for green-magenta (a*) and blue-yellow (b*) channels.

HSV colorspaces

HSV colorspaces are presented in the multi-value traits file as 1 to 359 for hue (H), and as 0 to 100 for saturation (S) and value (V).

 

Printing images from steps along the pipeline

If the command pcv.print_image() was used to generate images at certain steps along the pipeline, copies of these images should appear in your directory after the pipeline has been successfully run. Figure 3 shows examples of these commands.

You may choose to print copies of each image run through the pipeline that shows, for example, which objects were identified in the image and their seed assignments (Fig. 4, A), a visualization of geometric measures for each object (Fig. 4, B), or how pixels were classified with Naïve Bayes (Fig. 4, C).

You may also choose to print figures summarizing color data for each image (Fig. 5). 

 

Success! You have completed this How-To module. Return to Main Page.

Related articles

Filter by label

There are no items with the selected labels at this time.

Breeding Insight is funded by USDA at Cornell University. For more information, please visit http://www.breedinginsight.org.