Participate in a user-study regarding image quality degradations. You will be presented with 300 images, in the form of a browser-based study. Help us determine what types of quality degradations are shown in each image, and provide a quality rating.
We're looking for participants that are experienced with photography (or photo editing) and have a good understanding of the different types of quality degradations that online images often show. Please describe your level of experience with photography when making a bid for the project.
This experiment is part of a research project that we're conducting at University of Konstanz in Germany. Our group is working on image quality assessment. In this particular case we're looking into how well people can relate to the different types of degradations that are often encountered in public images available online e.g. photo sharing communities.
We are studying technical quality. In contrast to the aesthetic quality of an image, technical quality refers exclusively to the annoyance of different types of degradations that are visible in images. For instance, we're not concerned with the composition of the photograph, its framing or pleasantness of the subject material.
- create a [url removed, login to view] contributor account at [url removed, login to view]
- we'll invite you to participate in the study with the email address you've used to create your crowdflower account
- report your crowdflower email address and "worker id", so we can invite you to the study and confirm your participation after you've taken it
- complete the study, by answering at most three questions for each of the 300 images (see attached example for details)
If you have any questions feel free to ask.
The study takes about 1 hour to complete.
We need 5-10 participants in the study. As a consequence, I'll have to wait until enough bids are placed before accepting anyone, otherwise the bidding is closed automatically.
After completing the image quality study on crowdflower.com, please fill in the feedback form here: