Use Napari to draw a bounding box¶

Use an interactive Napari viewer to draw a rectangle that is converted to the appropriate parameters for either defining an ROI or bounding box for deep learning models like Segment Anything.

plantcv.annotate.napari_bbox(viewer, layername)

returns a list with x and y coordinates of the top left corner of the box, height and width. If more than one rectangle is drawn in the Napari shapes layer, the function returns a list of lists containing the parameters for each shape.

• Parameters:

  • viewer - a Napari viewer object. Should have a shapes layer with one or more rectangles drawn.
  • layername - the name of the shapes layer where the rectangles have been drawn.

• Context:

  • Used to convert easily drawn rectangles to [x, y, h, w], the format for either defining an ROI for image analysis with PlantCV or a bounding box for deep learning using Segment Anything.

• Example use below:

import plantcv.plantcv as pcv 
import plantcv.annotate as pcvan

# Open and image and a Napari viewer
img, _, _ = pcv.readimage("./exampleimage.png")

viewer = pcvan.napari_open(img=img)

# Add a shapes layer
viewer.add_shapes(name="shapes")

# In the interactive viewer, draw a rectangle. In the next cell, run:
boxes = napari_bbox(viewer=viewer, layername="shapes")

Source Code: Here