Unlocking Pre-trained Image Backbones for Semantic Image Synthesis

20 Dec 2023  ·  Tariq Berrada, Jakob Verbeek, Camille Couprie, Karteek Alahari ·

Semantic image synthesis, i.e., generating images from user-provided semantic label maps, is an important conditional image generation task as it allows to control both the content as well as the spatial layout of generated images. Although diffusion models have pushed the state of the art in generative image modeling, the iterative nature of their inference process makes them computationally demanding. Other approaches such as GANs are more efficient as they only need a single feed-forward pass for generation, but the image quality tends to suffer on large and diverse datasets. In this work, we propose a new class of GAN discriminators for semantic image synthesis that generates highly realistic images by exploiting feature backbone networks pre-trained for tasks such as image classification. We also introduce a new generator architecture with better context modeling and using cross-attention to inject noise into latent variables, leading to more diverse generated images. Our model, which we dub DP-SIMS, achieves state-of-the-art results in terms of image quality and consistency with the input label maps on ADE-20K, COCO-Stuff, and Cityscapes, surpassing recent diffusion models while requiring two orders of magnitude less compute for inference.

PDF Abstract
Task Dataset Model Metric Name Metric Value Global Rank Benchmark
Image-to-Image Translation ADE20K Labels-to-Photos DP-SIMS (ConvNext-L) mIoU 54.3 # 1
FID 22.7 # 1
Image-to-Image Translation Cityscapes Labels-to-Photo DP-SIMS (ConvNext-L) mIoU 76.3 # 1
FID 38.2 # 1
Image-to-Image Translation COCO-Stuff Labels-to-Photos DP-SIMS (ConvNext-XL) FID 13.3 # 1
Image-to-Image Translation COCO-Stuff Labels-to-Photos DP-SIMS (ConvNext-L) FID 13.6 # 2