CQFIO/PhotographicImageSynthesis

Health v2 · maintenance only

Flags: no_releases no_license

Adoption not part of the score

1248 stars · 222 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

A TensorFlow implementation of the ICCV 2017 paper 'Photographic Image Synthesis with Cascaded Refinement Networks', which synthesizes photographic images from semantic label maps using a feedforward network trained with direct regression. It includes pretrained models for 512p and 1024p synthesis and training scripts for progressive fine-tuning.

Use cases

• generate realistic images from semantic segmentation maps
• synthesize photographic street scenes from layout sketches
• render indoor scenes from semantic layouts
• train an image synthesis model at 256p and fine-tune to higher resolutions
• reproduce ICCV 2017 image synthesis research results

When to choose

• you need to generate photographic images conditioned on semantic layouts
• you want a non-GAN, regression-based image synthesis baseline
• you are reproducing or building on the cascaded refinement networks paper
• you need high-resolution (up to 2-megapixel) layout-to-image synthesis

When to avoid

• you need actively maintained code or modern framework support (TensorFlow >=1.0 era)
• you want GAN-based or diffusion-based image generation
• you need a production-ready image generation service
• you cannot access a CUDA GPU with sufficient memory for high-resolution synthesis

Facets

library · maturity maintenance

machine-learning deep-learning image-processing machine-learning computer-vision image-processing artificial-intelligence python image-synthesis semantic-layout cascaded-refinement-networks tensorflow generative-models research-code iccv-2017 linux gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="CQFIO/PhotographicImageSynthesis")