Visual Attribute Transfer through Deep Image Analogy
This addresses the challenge of visual attribute transfer for computer vision and graphics applications, offering a novel method but with incremental improvements over existing techniques.
The paper tackles the problem of transferring visual attributes like color and texture between images with different appearances but similar semantic structures, achieving effective results across various applications such as style transfer and sketch-to-photo conversion.
We propose a new technique for visual attribute transfer across images that may have very different appearance but have perceptually similar semantic structure. By visual attribute transfer, we mean transfer of visual information (such as color, tone, texture, and style) from one image to another. For example, one image could be that of a painting or a sketch while the other is a photo of a real scene, and both depict the same type of scene. Our technique finds semantically-meaningful dense correspondences between two input images. To accomplish this, it adapts the notion of "image analogy" with features extracted from a Deep Convolutional Neutral Network for matching; we call our technique Deep Image Analogy. A coarse-to-fine strategy is used to compute the nearest-neighbor field for generating the results. We validate the effectiveness of our proposed method in a variety of cases, including style/texture transfer, color/style swap, sketch/painting to photo, and time lapse.