Decompose Radicals, Then Reward: Fine-Grained Inspection for Accurate Chinese Text Rendering

Hugging Face Daily Papers Papers
text-to-image chinese-text-rendering reinforcement-learning ocr grpo multimodal

Summary

提出了基于表意文字描述序列(IDS)的IDSpect方法，通过细粒度的部件级与空间关系奖励信号改进中文文本渲染，GRPO后训练Qwen-Image在LongText和GenTextEval上取得领先表现。

Rendering accurate Chinese text remains challenging for text-to-image models. Existing OCR-based reinforcement-learning rewards compare decoded transcripts with target strings. Such rewards overlook the compositional nature of Chinese writing: an ideograph consists of reusable components arranged through explicit spatial relations, yet OCR evaluates it as an atomic character. Consequently, visually different radical-level errors may receive equally coarse feedback, encouraging glyphs that merely resemble the target instead of faithfully reproducing its internal structure. We employ Ideographic Description Sequences (IDS), which comprise spatial operators and character components, and train an expert IDS recognizer to transcribe rendered Chinese text into this representation. Building on this recognizer, we introduce IDSpect, which deterministically decomposes the target text into IDS tokens and aligns crop-level visual IDS predictions with the target sequence. Globally unique token credit makes this comparison robust to the order of detected text regions. Combined with a whole-character semantic reward, IDSpect supplies fine-grained credit with component and spatial-relation without changing the image generator or adding inference-time cost. Experiments with GRPO post-training of Qwen-Image demonstrate that IDSpect achieves leading structural quality and semantic alignment on LongText and GenTextEval.
Original Article
View Cached Full Text

Cached at: 10/01/26, 04:21 AM

Paper page - Decompose Radicals, Then Reward: Fine-Grained Inspection for Accurate Chinese Text Rendering

Source: https://huggingface.co/papers/2609.37569

Abstract

RenderingaccurateChinesetextremainschallengingfortext-to-imagemodels.ExistingOCR-basedreinforcement-learningrewardscomparedecodedtranscriptswithtargetstrings.SuchrewardsoverlookthecompositionalnatureofChinesewriting:anideographconsistsofreusablecomponentsarrangedthroughexplicitspatialrelations,yetOCRevaluatesitasanatomiccharacter.Consequently,visuallydifferentradical-levelerrorsmayreceiveequallycoarsefeedback,encouragingglyphsthatmerelyresemblethetargetinsteadoffaithfullyreproducingitsinternalstructure.WeemployIdeographicDescriptionSequences(IDS),whichcomprisespatialoperatorsandcharactercomponents,andtrainanexpertIDSrecognizertotranscriberenderedChinesetextintothisrepresentation.Buildingonthisrecognizer,weintroduceIDSpect,whichdeterministicallydecomposesthetargettextintoIDStokensandalignscrop-levelvisualIDSpredictionswiththetargetsequence.Globallyuniquetokencreditmakesthiscomparisonrobusttotheorderofdetectedtextregions.Combinedwithawhole-charactersemanticreward,IDSpectsuppliesfine-grainedcreditwithcomponentandspatial-relationwithoutchangingtheimagegeneratororaddinginference-timecost.ExperimentswithGRPOpost-trainingofQwen-ImagedemonstratethatIDSpectachievesleadingstructuralqualityandsemanticalignmentonLongTextandGenTextEval.

View arXiv pageView PDFAdd to collection

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2609.37569 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2609.37569 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2609.37569 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

TextVerifier: Seeing Beyond Recognition to Verify Chinese Text Rendering

Hugging Face Daily Papers

TextVerifier 提出了一种针对中文文本渲染缺陷的逐字符验证方法，利用表意文字描述序列（IDS）作为文本塔输入，并在 TextRenderBench-ZH 基准上达到 0.844 的 F1，显著优于 OCR 置信度、通用 VLM 和 TextPecker，还可作为 Flow-GRPO 后训练的逐字符奖励进一步降低渲染错误。

Efficient Scene-Text Refinement via Pixel-Space Grouped Image Generation

Hugging Face Daily Papers

PixText 将局部文本修正建模为像素空间的分组图像生成，仅生成目标区域并保留未编辑内容，借助 MeanFlow 实现四步快速生成，在新提出的 TextRefinementBench 上以 83.06% 的字形通过率大幅超越最强基线。

prunaai/p-image-ideogram

Replicate Explore

Pruna AI 发布的优化版文本生图模型 p-image-ideogram 已上线 Replicate，支持提示词理解、排版控制、思考级别调节与自定义分辨率，主打高保真写实与文字渲染能力。

UltraText Bench: A Comprehensive Bilingual Benchmark for Evaluating Visual Text Rendering in Image Generation

Hugging Face Daily Papers

UltraText Bench is a bilingual (English/Chinese) benchmark with 432 prompts across 24 scene categories for evaluating dense visual text rendering in image generation models, scored by the Q-Judge VLM across text fidelity, clarity, spatial quality, and scene quality — revealing strong trade-offs between model variants such as Z-Image-Turbo versus Z-Image-Base.