Enhance Your Photo With Interestingness Guided Img2Img Translation
Auf einen Blick
- Projektleiter/in : Prof. Dr. Helmut Grabner
- Co-Projektleiter/in : Prof. Dr. Wibke Weber
- Projektteam : Fitim Abdullahu, Valery Wyss
- Projektvolumen : CHF 20'000
- Projektstatus : laufend
- Drittmittelgeber : Interne Förderung (ZHAW digital / Digital Futures Fund)
- Kontaktperson : Helmut Grabner
Beschreibung
Images have the ability to capture and hold an individual's attention, which is the definition of visual interest. We distinguish between subjective and common interests. To maximise attention on social media and websites, you need to decide which images to use to generate the most common interest. First, we will develop an open web-based tool. Our computational model ranks images according to their level of interest (from common to subjective). The user can upload up to 10 images and will receive a ranking list. Second, we will use generative AI img2img (image-to-image) techniques to slightly modify but
significantly enhance and refine the attractiveness of the images in question. Again, the user can upload an image and the tool will suggest alternative images, which will then be made available for download. The goal is to help users and professionals to post images of common interest, thus attracting higher attention through impactful stories and posts.