Research map / Image generation / Autoregressive and latent generation
Deep Markov Random Field for Image Modeling
Zhirong Wu, Dahua Lin, Xiaoou Tang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use Deep MRF instead of the CNN super-resolution baseline | super-resolution · Set5 x3 upscale | PSNR (dB): 32.30 -> 33.14 | Improved |
| use Deep MRF instead of the CNN baseline | super-resolution · Set5 x4 upscale | PSNR (dB): 30.09 -> 30.52 | Improved |
| use Deep MRF instead of the CNN baseline | super-resolution · Set5 x2 upscale | PSNR (dB): 36.34 -> 36.38 | No clear effect |
| use Deep MRF instead of the CNN baseline | super-resolution · Set14 x3 upscale | PSNR (dB): 29.00 -> 29.38 | Improved |
| use Deep MRF instead of the CNN baseline | super-resolution · BSD100 x3 upscale | PSNR (dB): 28.20 -> 28.54 | Improved |
| use Deep MRF instead of CSCN | super-resolution · Set14 x3 upscale | PSNR (dB): 29.41 -> 29.38 | Hurt |
| use Deep MRF instead of SE on the hard butterfly image | super-resolution · Set5 butterfly x3 upscale | PSNR (dB): 26.86 -> 29.09 | Improved |
| use Deep MRF instead of SE on baby image | super-resolution · Set5 baby x2 upscale | PSNR (dB): 38.48 -> 38.31 | Hurt |
| increase training patch size (ablation) | Deep MRF texture model · Brodatz/VisTex texture synthesis | visual quality | Improved |
| increase number of Gaussian mixtures (ablation) | Deep MRF texture model · Brodatz/VisTex texture synthesis | visual quality | Improved |
| add backward pass (coupled opposite-direction RNNs) over a single raster-order 2D-RNN (ablation) | Deep MRF · texture synthesis | visual quality | Improved |
Closest papers by meaning
- Neural Markov Random Field for Stereo Matching (CVPR 2024 · similarity 0.90)
- Combining Markov Random Fields and Convolutional Neural Networks for Image Synthesis (CVPR 2016 · similarity 0.90)
- CNN in MRF: Video Object Segmentation via Inference in a CNN-Based Higher-Order Spatio-Temporal MRF (CVPR 2018 · similarity 0.88)
- A Discriminative View of MRF Pre-Processing Algorithms (ICCV 2017 · similarity 0.87)
- MarkovGen: Structured Prediction for Efficient Text-to-Image Generation (CVPR 2024 · similarity 0.85)
- How to Train Neural Field Representations: A Comprehensive Study and Benchmark (CVPR 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
