A Glance Is All You Need: Single-Pass Fine-Grained Image Captioning with SimLoss

• 类型：arxiv
• 标识：2609.00591
• 链接：https://arxiv.org/abs/2609.00591
• 主分类：rag
• 形态：position
• TLDR：An image may be worth a thousand words, but most captioning models describe it in only a few. Modern vision-language models produce fluent high-level captions, yet routinely miss the attributes, counts, textures, materials, and spatial relations that make an image visually specific. Recent multi-stage systems recover some of these details through generation, decomposition, verification, and rewriting, but they do so at the expense of substantially higher inference latency. We propose SimLoss, a reference-free embedding-space objective for single-pass fine-grained image captioning. SimLoss trai
• 待LLM分类：否
• 来源文件：
• /inbox/tom/_candidates/2026-09-03-agent-rag-longcontext-candidates.json