The ImageHelpfulnessMetric is a multimodal metric designed to evaluate the helpfulness of images in enhancing the understanding of textual content. It assesses how well the image supports the text, serving as a proxy for evaluating the performance of models that generate or select images to complement text.

​
Required Arguments

  • input: A list containing the textual context or description.
  • actual_output: A list containing the textual description of the image and the image itself. The image is represented as an MLLMImage instance with:
  • url: The file path or URL to the image.

​
Optional Arguments

  • threshold: A float representing the minimum passing threshold, defaulted to 0.5.
  • model: A string specifying which of OpenAI’s GPT models to use, or any custom LLM model of type DeepEvalBaseLLM. Defaulted to ‘gpt-4o’.
  • include_reason: A boolean which, when set to True, includes a reason for its evaluation score. Defaulted to True.
  • strict_mode: A boolean which, when set to True, enforces a binary metric score: 1 for perfection, 0 otherwise. It also overrides the current threshold and sets it to 1. Defaulted to False.
  • async_mode: A boolean which, when set to True, enables concurrent execution within the measure() method. Defaulted to True.
  • verbose_mode: A boolean which, when set to True, prints the intermediate steps used to calculate the metric to the console. Defaulted to False.

​
Usage Example

import sys
import os

from agensight.eval.metrics import ImageHelpfulnessMetric
from agensight.eval.test_case import MLLMTestCase
from agensight.eval.test_case import MLLMImage

input_data = ["This is a report with an image."]
actual_output = ["The following image shows a cat on a windowsill.", MLLMImage(url="image-path/cat.jpeg")]

metric = ImageHelpfulnessMetric(model="gpt-4o", threshold=0.5)
test_case = MLLMTestCase(input=input_data, actual_output=actual_output)

metric.measure(test_case)
print(f"Score: {metric.score}")
print(f"Reason: {metric.reason}")