Label Image with Napari

This function opens an image in Napari and then defines a set of Points layers with the user-defined labels called classes. A random shape of the annotation symbol is assigned to each of the classes. Image can be annotated as long as viewer is open.

plantcv.annotate.napari_label_classes(img, classes, size=10, importdata=False, show=True)

returns napari viewer object

  • Parameters:

    • img - image data (compatible with gray, RGB, and hyperspectral data. If data is hyperspecral it should be the array e.g. hyperspectral.array_data)
    • classes - list of classes to label. This option is not necessary if data is data is imported.
    • size - integer pixel size of label (also adjustable from the interactive Napari viewer)
    • importdata - dictionary of data, data saved from napari_save_coor or data imported from napari_read_coor
    • show - if show=True, viewer is launched. False setting is useful for test purposes.

  • Context:

    • Adding one or more classes of points layer for annotation of the image.

  • Example use:

    • Ground truth counting, labeling classes of objects of interest.
import plantcv.plantcv as pcv 
import plantcv.annotate as pcvan
import napari

# Create an instance of the Points class
img, path, name = pcv.readimage("./grayimg.png")

# Opens interactive napari viewer
viewer = pcvan.napari_label_classes(img=img, classes=['background', 'wing','seed'], size=30)

Screenshot

Source Code: Here