Welcome
Thank you for joining our research! Your involvement will help to improve automatically-generated image captions and assit blind and visually impaired people learn about image contents.
To guarantee the quality of data collected, please read carefully the following instructions and pass a test for getting started with the HIT.
Instruction
Guidance
Task 1
1. Look at the image.
2. You will be shown 5 captions. Please highlight the words/phrases you think are most important to understand the image and hope to appear in the caption.
3. You need to highlight 5 words/phrases. The words/phrases highlighted should have substantial & single meanings, such as "dog", "happy", "run" etc. instead of "at/in/on" "a/an/the" or "kind of".
4. The words/phrases highlighted need to be complete.
5. Click EXAMPLES below for reference.
Task 2
1. Rank the elements according to their importance in your opinion.
2. From 1 to 5 it indicates "the most important" to "the least important"
Task 3
1. Choose your favorite caption.
2. Input reasons for your selections above.
3. Click Submit button and done!
Operation
- Click and drag to the end of words/phrases to highlight the elements you think are important.
- You can change your selection by clicking the undo last step button to withdraw your last move, or the clear all button to redo the test.
Right example
|
Wrong example
|
