Visual attributes used by Artificial Intelligence models to make predictions


In this HIT, we will show you two different visualizations related to a visual attribute. Your task is to determine whether the visual attribute is the same in both the visualizations or not. Occasionally, the images may contain disturbing or adult content. We would like to remind you that participation in our study is voluntary and that you can withdraw from the study at any time.

  • Images
    a set of five images for which the model detects the visual attribute with high scores. By looking at what these images have in common, you can try to understand the meaning of the visual attribute.
  • Highlighted visual attribute
    for the same set of five images, we show a heatmap of where the model is looking at when detecting the visual attribute. The higher the intensity of the red color, the more relevant is that part of the image for the visual attribute. The higher the intensity of the blue color, the less relevant is that part of the image for the visual attribute. Example (water), Example (fingers)
 


Section A
Images
Highlighted visual
attributes


     

Section B
Images
Highlighted visual
attributes




Please answer the questions below

Q1. Which of the following best describes the focus of the red region in the Highlighted visual attributes?

               In both Section A and Section B, focus is on the same visual attribute.
                   

               In Section B, focus is on different visual attribute compared to Section A.
                   

               The focus of red region is unclear in Section A.
                   

               The focus of red region is unclear in Section B.
                   


Q2. How confident are you about your answer to Q1?
Q3. Describe the visual attributes in Section A and Section B for your answer to Q1? This question is mandatory for completing this HIT. Detailed answers will receive a bonus.