Projects
Distramera
Undergraduate Special Problem
Distramera is the output of my special problem entitled "A Novel Approach to Aligning Photographer Intention and Viewer Attention: Mobile Photography Assistance for iOS Devices using Saliency Map Prediction".
For this paper, I wanted to reframe photography not as following a bunch of rules, but instead as an optimization where you minimize saliency away from distactors and maximize the saliency of your subject(s). I think what's funny about this one is that this one uses machine learning. But, ho-ho no, it doesn't do the work for you, mind. Instead, it just tells you where in the photo is the most salient! That's it. Creative control is still up to the photographer.
If you would look at the first photo, the photographer chose the drain as the intended subject (highlighted by the yellow box). But the application has detected that the blue slippers are a distraction from the intended subject. In other words, the intended subject and the most salient subject are not the same. If instead you chose the blue slippers as the intended subject, it will say that the intended subject and the most salient subject are one and the same.
I have left this paper unpublished as I thought that I could do better. And it will always be a reminder that I could do better than this!
To request access to the paper or other materials including the demo video, you can contact the Institute of Computer Science of the University of the Philippines Los Baños.
Tech Stack
- SwiftUI
- UIKit
- AVFoundation
- Metal Core Image (for image processing)