Inspiration
A picture/video is too pale without any enhancer. It's also boring if we always use the same filter for all the pixels. We want a clever way!
What it does
Firstly, it parsers the pictures by marking the objects in it. Then it applies different methods based on the categories of the objects, such as: blur on background, sharpen on people, 'magic effects' on moving objects. It can also enhance videos!
How I built it
We use DeepLab to understand the pictures. We use openCV to process images. We use VGG-16 to transfer the styles of a picture.
Challenges I ran into
It's very very slow to run and test models with CPU.
Accomplishments that I'm proud of
We have something that makes my friends laugh a lot.
What I learned
Knowledge of image procession, a better understanding of (deep) neural networks in the fields of image recognition
What's next for AutoEnhancerrrrrrr
Beg for tensorflow TPU.
