Github: https://github.com/woozzu/tagan
Task: manipulating images using natural language description
semantically modify visual attributes of an object in an image according to the text describing the new visual appearance
a sample:
existing methods:
they do not fully preserve text-irrelevant contents of the original image
a sample: background changed
Our method:
The key to our method is the text-adaptive discriminator that creates word-level local discriminators according to
input text to classify fine-grained attributes independently.With this discriminator, the generator learns to generate images where only regions that correspond to the
given text are modified.
Related task: image-to-image translation, text-to-image synthesis
Network:
每个词都有个local discriminator, 来判断是否与原图relevant.
卖点：之前没人考虑做保持背景的只改变visual attribute的任务，算是第一人，效果对比【惊艳】。
缺点：没有量化指标，只能靠User Study 和主观判断
Loss:
主要靠reconstruction loss来控制背景的不变
引用自《Arbitrary facial attribute editing: Only change what you want》 submmited to TIP