对比两个视觉语言模型架构，指导人体姿态检测系统设计
An Architecture-Led Hybrid Report on Body Language Detection Project
- 基于模型架构分析选择适合的视觉语言模型
- 实现视频帧到标注框的自动化生成与结构校验
- 适合关注多模态系统工程与输出可靠性研究者
本报告从架构视角分析两种现代视觉语言模型（VLMs）Qwen2.5-VL-7B-Instruct 和 Llama-4-Scout-17B-16E-Instruct，阐释其架构特性如何映射至 BodyLanguageDetection 仓库中实现的视频转成果物流水线。系统通过采样视频帧，调用 VLM 检测可见人物并生成带提示条件属性（默认为情绪）的像素空间边界框，利用预定义模式验证输出结构，并可选地渲染带注释视频。首先总结两者共有的多模态基础（视觉分词、Transformer 注意力、指令遵循），再以足够深度描述各自架构以支撑工程决策，不涉及推测性内部机制。最后将模型行为与系统约束关联：结构化输出可能语法正确但语义错误，模式验证仅限结构而非几何正确性，人物标识符在当前提示契约下为帧内局部，交互式单帧分析返回自由文本而非受模式约束的 JSON。这些差异对撰写可信结论、设计鲁棒接口和规划评估至关重要。
原文摘要 · Abstract (English)
This report provides an architecture-led analysis of two modern vision-language models (VLMs), Qwen2.5-VL-7B-Instruct and Llama-4-Scout-17B-16E-Instruct, and explains how their architectural properties map to a practical video-to-artifact pipeline implemented in the BodyLanguageDetection repository [1]. The system samples video frames, prompts a VLM to detect visible people and generate pixel-space bounding boxes with prompt-conditioned attributes (emotion by default), validates output structure using a predefined schema, and optionally renders an annotated video. We first summarize the shared multimodal foundation (visual tokenization, Transformer attention, and instruction following), then describe each architecture at a level sufficient to justify engineering choices without speculative internals. Finally, we connect model behavior to system constraints: structured outputs can be syntactically valid while semantically incorrect, schema validation is structural (not geometric correctness), person identifiers are frame-local in the current prompting contract, and interactive single-frame analysis returns free-form text rather than schema-enforced JSON. These distinctions are critical for writing defensible claims, designing robust interfaces, and planning evaluation.
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。