log_image

Log a matplotlib Figure, PIL Image, or path to a png/jpg image to Modelbit.

The image will show in the logs viewer of the deployment or training job.

Parameters

mb.log_image({image})

• image: [matplotlib.pyplot.Figure | PIL.Image | str] This argument can be a matplotlib Figure, a PIL Image, or a string filepath to a local .png or .jpg image.

Returns

None

Outside of a Modelbit deployment, calling log_image prints a success message if it's able to process the supplied image. If the image argument cannot be processed for logging an exception is raised.

Example

Logging a matplotlib Figure

When logging a matploylib image, pass the figure to log_image:

import matplotlib.pyplot as plt
from random import random

def example_plot():
  fig, ax = plt.subplots()
  ax.plot([1, 2, 3, 4], [random(), random(), random(), random()])
  mb.log_image(fig)
  return { "prediction": 1 }

Logging the default matplotlib Figure

When you run plt.plot and related API calls directly rather than on subplots, pass plt.gcf() as the argument to mb.log_image:

import matplotlib.pyplot as plt

def example_plot():
  plt.plot([1, 2, 3, 4], [random(), random(), random(), random()])
  mb.log_image(plt.gcf())
  return { "prediction": 1 }

Logging a PIL Image

Pass an Image instance when using PIL:

from PIL import Image

def example_plot():
  img = Image.open("my_image.jpg")
  mb.log_image(img)
  return { "prediction": 1 }

Logging an image file

When your image is stored locally as a png or jpg, pass the path to that file to log_image:

def example_plot():
  mb.log_image("my_image.jpg")
  return { "prediction": 1 }

Logging multiple images

To log multiple images, call log_image multiple times:

import matplotlib.pyplot as plt

def example_plot():
  fig, ax = plt.subplots()

  print("Before:")
  ax.plot([1, 2, 3, 4], [random(), random(), random(), random()])
  mb.log_image(fig)

  print("After:")
  ax.plot([1, 2, 3, 4], [random(), random(), random(), random()])
  mb.log_image(fig)

  return { "prediction": 1 }