Research map / Image generation / Autoregressive and latent generation
Region-Native Visual Tokenization
Mengyu Wang, Yuyao Huang, Henghui Ding, Xinlong Wang, Tiejun Huang, Yao Zhao, Yunchao Wei, Shuicheng Yan
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 arbitrarily shaped region tokens instead of grid tokens | Reader · ImageNet val (reconstruction) | rFID: 5.24 -> 2.43 (vs RQ-VAE(16384)) | Improved |
| replace region tokens of mouth and lips with those from a smiling target image | Reader · Smile transferring (CelebAMask-HQ) | ID Score: 0.7452 -> 0.8470 (vs EditGAN+30) | Improved |
| region-token editing versus EditGAN on attribute accuracy | Reader · Smile transferring (CelebAMask-HQ) | Attribute accuracy (%): 91.5 -> 84.2 (EditGAN vs Reader) | Hurt |
| keep only pseudo-labels with area proportion under 1/16 for training (ablation) | Reader · ImageNet val, region area proportion | L1 distance: 1.44, 1.56 below 1/16 vs 2.53, 3.46 above | Improved |
| use ViT patch size 8x8 with bidirectional block and masked attention (ablation) | Reader · ImageNet val (reconstruction) | rFID | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 3 on most shared results.
- EditGAN — EditGAN: High-Precision Semantic Image Editing (NeurIPS 2021) · better on 1 of 2 shared results
- InterFaceGAN — InterFaceGAN: Interpreting the Disentangled Face Representation Learned by GANs (TPAMI 2020) · better on 1 of 1 shared result
- MaskGAN — MaskGAN: Better Text Generation via Filling in the _______ (ICLR 2018) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- TokenFlow: Unified Image Tokenizer for Multimodal Understanding and Generation (CVPR 2025 · similarity 0.87)
- From Pixels to Tokens: Byte-Pair Encoding on Quantized Visual Modalities (ICLR 2025 · similarity 0.87)
- Image Understanding Makes for A Good Tokenizer for Image Generation (NeurIPS 2024 · similarity 0.86)
- ENAT: Rethinking Spatial-temporal Interactions in Token-based Image Synthesis (NeurIPS 2024 · similarity 0.86)
- Channel-wise Vector Quantization (NeurIPS 2026 · similarity 0.86)
- UniFlow: A Unified Pixel Flow Tokenizer for Visual Understanding and Generation (ICLR 2026 · similarity 0.86)
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.
