原生关联：基于基础视觉语言模型的野外置信度感知人类感知
Native Association: Confidence-Aware Human Perception in the Wild with a Foundation VLM
浏览论文内容
中文总结 AI 辅助
本研究提出原生关联方法，通过微调0.77B的Florence-2模型，以语法约束序列直接输出球员属性，消除身份互换，相比零样本API将错误关联降低4倍，并实现检测F1达0.95，支持置信度拒绝与缩放重读，表明增益源于结构而非权重。
中文摘要 AI 辅助
从广播画面中提取谁在何处、属于哪支队伍、身穿哪个号码，通常是通过拼接检测器、OCR引擎和分类器来实现的——而拼接步骤在遮挡情况下会导致身份互换。我们转而让关联成为原生操作：一个0.77B参数的视觉语言模型（Florence-2）经过微调，以语法约束的序列形式输出每个人的所有属性，每个属性在其所属者的块内生成。因此，输出在每一帧上天然符合模式，且不存在事后绑定步骤将正确读取的号码附加到错误的球员身上：残余的错误关联纯粹是感知错误，其发生概率比零样本提示的前沿API低约4倍（0.057对比0.21-0.24）。在冻结的多运动测试集上，这一单次通过达到了0.95的检测F1分数（API为0.65-0.75）。一次额外的前向传递即可获得每个字段的置信度，支持拒绝选项（在50%覆盖率下，球衣精度从0.71提升至0.96），并为小球员路由无训练的缩放重读。令人惊讶的是，一旦语法被学习，在我们测试的适应配置下，进一步的参数高效调优并未带来可衡量的增益；相同的配方在WIDER-Attribute上达到了93.1的给定框mAP，在其标准测试协议下首次产生了检测耦合的端到端结果（84.5 mAP），并复现了相同的调优结果。在这种机制下，增益存在于结构中，而非增加的权重中。
英文摘要
Extracting who is where, on which team, wearing which number from a broadcast frame is typically done by stitching a detector, an OCR engine, and classifiers together -- and the stitching step swaps identities under occlusion. We make association native instead: a 0.77B vision-language model (Florence-2) is fine-tuned to emit all per-person attributes as one grammar-constrained sequence, with each attribute generated inside its owner's block. Output is therefore schema-valid on every frame by construction, and no post-hoc binding step exists to attach a correctly read number to the wrong player: residual misassociation is pure perception error, $\approx4\times$ rarer than zero-shot-prompted frontier APIs' (0.057 vs. 0.21-0.24). On a frozen multi-sport test set, this single pass reaches 0.95 detection F1 (APIs: 0.65-0.75). A single extra forward pass yields a per-field confidence that supports a reject option (jersey precision $0.71\rightarrow0.96$ at half coverage) and routes a training-free zoom-and-re-read for small players. Surprisingly, once the grammar is learned, further parameter-efficient tuning yields no measurable gain under the adaptation configurations we test; the identical recipe on WIDER-Attribute reaches 93.1 mAP given-box, yields the first detection-coupled end-to-end results under its standard test protocol (84.5 mAP), and reproduces the same tuning result. In this regime, the gains live in the structure, not in added weights.
发表机构
- WSC Sports（WSC体育公司(或WSC Sports，按原文保留)）
机构由 AI 辅助整理，请以论文原文为准。
