隐私优先：边缘部署Privacy-First: Edge Deployment
模型基于ONNX格式，在手机/浏览器本地运行，舌象照片不上传服务器。你的健康数据就是你的私钥。Model uses ONNX format, running locally on phone/browser. Tongue photos never leave your device. Your health data = your private key.
中医"望闻问切"中，望舌是最直观、最易于数字化的诊断方法。舌象的变化——舌色、舌苔、舌形——可能反映人体内部的健康趋势。但长期以来，这一方法受限于两个瓶颈：一是需要经验丰富的医师，二是缺乏客观量化标准。Among TCM's "four diagnostic methods," tongue observation is the most intuitive and easiest to digitize. Changes in tongue appearance—color, coating, shape—may reflect internal health trends. But this method has been limited by two bottlenecks: the need for experienced practitioners and the lack of objective quantification standards.
我们开源了舌象分析AI模型，核心设计理念：We've open-sourced a tongue analysis AI model with core design principles:
模型基于ONNX格式，在手机/浏览器本地运行，舌象照片不上传服务器。你的健康数据就是你的私钥。Model uses ONNX format, running locally on phone/browser. Tongue photos never leave your device. Your health data = your private key.
结合舌色分析、舌苔厚度检测、舌形轮廓识别，多维度信息融合可能有助于更全面地理解健康趋势。Combining tongue color analysis, coating thickness detection, and shape contour recognition. Multi-dimensional information fusion may help with more comprehensive health trend understanding.
代码在GitHub上完全开放，任何人可以审计、改进、部署。没有黑箱，没有数据垄断。Code is fully open on GitHub. Anyone can audit, improve, and deploy. No black boxes, no data monopolies.
我们刻意避免使用"诊断""治疗"等医疗敏感词。舌象辨识提供的是健康趋势的参考信息——就像天气预报告诉你明天可能下雨，但最终是否带伞还是你自己决定。这就是健康主权：信息透明、决策自主、数据自控。We deliberately avoid medical terms like "diagnosis" or "treatment." Tongue analysis provides reference information about health trends—like a weather forecast telling you it might rain tomorrow, but you decide whether to carry an umbrella. This is health sovereignty: transparent information, autonomous decisions, self-controlled data.