医学影像理解双任务竞赛中，多模型融合与零样本推理实现高精度概念识别与图文生成。
DS@GT ARC at ImageCLEFmedical 2026: Architectural Diversity for Concept Detection and Foundation-Model Scaling for Caption Prediction in Medical Image Analysis
- 三模型晚融合+正则化阈值调优，避免罕见概念过拟合
- 主任务F1达0.5790，次任务F1达0.9657，性能领先
- 支持从零训练到全微调的多种模型规模方案，兼顾成本与效果
本文介绍DS@GT团队在ImageCLEFmedical Caption 2026挑战赛中的提交方案，该挑战延续对ROCOv2数据集的长期基准评测，包含两个任务：概念检测（任务1）为放射科图像分配UMLS概念唯一标识符（CUIs），以及标题生成（任务2）生成自然语言描述。任务1的主提交采用ConvNeXt-V2、BiomedCLIP ViT-B/16与DenseNet-169三模型晚融合集成，并引入正则化“诚实阈值调优”以防止稀有概念上的验证过拟合；该方案在官方提交中取得主指标F1=0.5790，次指标F1=0.9657，排名第一。同时提交的无训练KNN检索管道基于冻结的BiomedCLIP嵌入，获得主指标F1=0.5780，次指标F1=0.9599，成本极低且性能接近微调集成。任务2涵盖微调的Gemma-3 27B模型（总体F1=0.3571，排名第三）、完全微调的BLIP流程（含自定义Vizwins融合，F1=0.3564）以及零样本MedGemma-4B结合PubMed风格提示（F1=0.3186），覆盖从大模型到小模型的多尺度方案。代码已开源：https://github.com/dsgt-arc/imageclef-caption-2026。
原文摘要 · Abstract (English)
We describe the DS@GT submissions to the ImageCLEFmedical Caption 2026 challenge, which continues a long-running benchmark on the ROCOv2 dataset with two tracks: Concept Detection (Task 1), assigning UMLS Concept Unique Identifiers (CUIs) to radiology images, and Caption Prediction (Task 2), generating natural-language captions. For Task 1, our primary submission was a three-way late-fusion ensemble of ConvNeXt-V2, BiomedCLIP ViT-B/16, and DenseNet-169 with a regularized ''Honest Threshold Tuning'' procedure designed to avoid validation overfitting on rare concepts; this submission ranked first on the official submission with a primary $F_1$ of $0.5790$ and a secondary $F_1$ of $0.9657$. In parallel, we submitted a training-free KNN retrieval pipeline over frozen BiomedCLIP embeddings, which reached a primary $F_1$ of $0.5780$ and a secondary $F_1$ of $0.9599$-essentially matching the fine-tuned ensemble on the primary track at a fraction of the cost. For Task 2, our submissions included a fine-tuned Gemma-3 27B model (overall $0.3571$, ranking third in the official submission), a fully fine-tuned BLIP pipeline with custom Vizwins merging ($0.3564$), and a zero-shot MedGemma-4B run with a PubMed-style prompt ($0.3186$), spanning a wide range of model scales and training costs. Code: https://github.com/dsgt-arc/imageclef-caption-2026.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。