MixNMatch: Multifactor Disentanglement and Encoding for Conditional Image Generation

TL;DR

MixNMatch is a novel conditional generative model that disentangles and encodes multiple image factors like background, pose, shape, and texture with minimal supervision, enabling versatile mix-and-match image generation.

Contribution

It extends FineGAN by incorporating factor disentanglement and encoding for conditional generation with minimal supervision, using adversarial training and bounding boxes.

Findings

Successfully disentangles multiple image factors

Enables diverse applications like sketch2color and img2gif

Achieves accurate factor encoding and combination

Abstract

We present MixNMatch, a conditional generative model that learns to disentangle and encode background, object pose, shape, and texture from real images with minimal supervision, for mix-and-match image generation. We build upon FineGAN, an unconditional generative model, to learn the desired disentanglement and image generator, and leverage adversarial joint image-code distribution matching to learn the latent factor encoders. MixNMatch requires bounding boxes during training to model background, but requires no other supervision. Through extensive experiments, we demonstrate MixNMatch's ability to accurately disentangle, encode, and combine multiple factors for mix-and-match image generation, including sketch2color, cartoon2img, and img2gif applications. Our code/models/demo can be found at https://github.com/Yuheng-Li/MixNMatch