Unverified50% confidenceTradeoffExact time
页面状态表征主要有两种实现路径：基于DOM解析和基于视觉的方式；DOM解析精确度高、token消耗少但对动态渲染页面适应性弱，视觉方式更接近人类操作但计算成本更高
1
Sources
50%
Confidence
Long-term
Relevance
8/19/2026
First Seen
Sources
Related Claims
VerifiedBrowser Use offers two page-understanding modes: DOM mode and visual mode.80% similarUnverifiedHybrid approaches are becoming a trend: prioritizing DOM information and falling back to visual recognition when the DOM is unavailable or unreliable.77% similarUnverifiedBrowser Use等大模型驱动工具在每一步操作时需要将当前页面的DOM结构或截图发送给大模型进行推理，一个典型网页DOM序列化后可能超过数万Token76% similarUnverifiedCompared to traditional tools based on DOM tree parsing (such as Selenium and Playwright), screenshot-based perception can handle elements rendered via Canvas, WebGL, and other means inaccessible through the DOM, but is sensitive to page layout changes and introduces additional latency and token consumption.75% similarUnverifiedPage-Agent 选择直接解析 DOM 树的路线，以结构化数据替代视觉感知72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/773122API
curl https://kongchang.com/api/v1/knowledge/claims/773122MCP
get_claim(id=773122)