Table 2:

Improved image caption rating scale.

r	Meaning
5	Objects, a general scene, and actions are correctly identified if present in the image. The caption describes what is seen and where things are in space. Interpretation of overall setting and/or event is included.
4	Objects and/or a general scene and/or an action are correctly identified but not every element is completely identified. The caption describes what is seen and where things are in space. There is no interpretation of an event.
3	Relevant objects are correctly identified. The caption describes what is seen but not where objects are in space. There is no description of the overall setting and no interpretation of an event.
2	Objects are partially correctly identified with some errors, but the caption is accurate enough to give an idea of what is happening in the image. The caption identifies most of the objects but might not identify everything. There is no interpretation of what anything means.
1	Objects are incorrectly identified. The caption gives the wrong idea about what is happening in the image.