解决文生图生成中的身份危机
Resolving the Identity Crisis in Text-to-Image Generation
用 GRPO 加组合式奖励（人脸相似度惩罚、跨样本身份重复惩罚、人数计数约束）微调 flow-matching 文生图模型，为多人物生成中的身份坍缩提供了一套无需真实数据的可复用后训练范式。
原文依据 DisCo fine-tunes flow-matching models using Group-Relative Policy Optimization (GRPO), guided by a compositional reward
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论最优的文生图模型在生成包含多个人物的场景时，会持续出现身份危机：产生重复的人脸、合并不同身份以及数错人数。我们提出 DisCo（带多样性约束的强化学习），这是一个直接优化图像内部以及生成样本组之间身份多样性的强化学习框架。DisCo 使用组相对策略优化（GRPO）微调流匹配模型，并由一个组合式奖励引导，该奖励包括：(i) 惩罚图像内部的人脸相似性，(ii) 抑制样本之间的身份重复，(iii) 强制人数准确，(iv) 通过人类偏好分数保持视觉保真度与提示词对齐。单阶段课程在提示词复杂度增加时稳定训练。重要的是，该方法不需要任何真实数据。在 DiverseHumans 测试集上，DisCo 达到 98.6% 的唯一人脸准确率和近乎完美的全局身份分布，在保持感知质量的同时优于开源与专有模型（如 Gemini、GPT-Image）。我们的结果确立了跨样本多样性是解决身份坍缩的关键维度，并将 DisCo 定位为一种可扩展、无需标注的多人物图像合成方案。项目主页：https://qualcomm-ai-research.github.io/disco/。
英文原文摘要 1357 字符
State-of-the-art text-to-image models suffer from a persistent identity crisis when generating scenes with multiple humans: producing duplicate faces, merging identities, and miscounting individuals. We present DisCo (Reinforcement with Diversity Constraints), a reinforcement learning framework that directly optimizes identity diversity both within images and across groups of generated samples. DisCo fine-tunes flow-matching models using Group-Relative Policy Optimization (GRPO), guided by a compositional reward that: (i) penalizes facial similarity within images, (ii) discourages identity repetition across samples, (iii) enforces accurate person counts, and (iv) preserves visual fidelity and prompt alignment via human preference scores. A single-stage curriculum stabilizes training as prompt complexity increases. Importantly, this method does not require any real data. On the DiverseHumans Testset, DisCo achieves 98.6% Unique Face Accuracy and near-perfect Global Identity Spread, outperforming open-source and proprietary models (e.g., Gemini, GPT-Image) while maintaining perceptual quality. Our results establish cross-sample diversity as a critical axis for resolving identity collapse, positioning DisCo as a scalable, annotation-free solution for multi-human image synthesis. Project page: https://qualcomm-ai-research.github.io/disco/.