上传一张正脸自拍，数秒内即可精准识别您的脸型，并获得专属发型及眼镜搭配推荐，支持生成报告下载。
最佳检测建议：正对镜头、光线均匀柔和、表情自然放松、避免头发遮挡额头。
或拖拽文件至此处 · 支持 JPG, PNG, WebP
仅需三个简单步骤，发现您的真实脸型与美学穿搭建议。
选择一张无修图、五官清晰露出的正脸照片。
MediaPipe 算法快速测量下颌骨、额头和颧骨长宽比例。
实时得出脸型归类、置信度评分及发型饰品推荐。
摒弃主观臆测，利用现代计算机视觉算法与面部骨骼几何学公式，得出客观精准的脸型结论。
照片不离设备，100% 在本地浏览器计算。
基于 WebAssembly 引擎，瞬间完成数万次矩阵运算。
无广告弹窗干扰，无需输入手机号或邮箱。
了解各类脸型的骨骼特征，轻松找到最适合自己的造型。
脸部长度与宽度接近，面颊圆润，下颌线条柔和。适合能增加顶部高度或纵向延伸感的发型。
最理想的脸型之一，比例匀称自然，下巴微微收拢，几乎能驾驭所有发型与镜框。
下颌角清晰硬朗，额头与下颌宽度相当，兼具英气与高级感的高辨识度轮廓。
额头与颧骨较宽，向尖翘的下巴逐渐收窄，具有灵动俏丽的少女感与上镜优势。
颧骨高耸立体，额头与下巴相对较窄，极具时尚超模感与雕塑美感。
脸长明显大于脸宽，侧脸线条流畅，给人知性成熟、干练优雅的第一印象。
Traditional face shape quizzes ask you to eyeball your own mirror reflection or hold a tape measure across your cheeks — a process that produces inconsistent guesses because most faces share traits from two adjacent categories. FaceAgeMeter replaces subjective guesswork with Google MediaPipe Face Landmarker, a real-time 3D surface mesh model that maps 478 three-dimensional facial coordinates directly inside your web browser. (Whereas our Age, Symmetry, and Emotion tools use a 68-point 2D landmark detector via face-api.js, face shape classification benefits from MediaPipe's denser 478-point contour mesh along the forehead hairline zone, temples, zygomatic arches, and lower mandible.)
Once the 478 landmarks are plotted, our geometric classifier measures four primary craniofacial ratios: Forehead Width (distance between the left and right temporal landmarks), Cheekbone Width (bizygomatic breadth across the widest outer malar points), Jawline Width & Angle (bigonial distance and slope from the jaw corner to the chin pogonion), and Vertical Face Length (midline distance from the upper forehead landmark to the bottom of the chin). By comparing your length-to-width ratio against the taper of your forehead and jaw, the algorithm classifies your structure into Oval, Round, Square, Heart, Diamond, or Oblong and computes a proportional confidence score.
Human faces exist on a continuous spectrum rather than six rigid boxes. Many people have a blended structure (such as an Oval-Heart or Soft Square). Camera focal length, beard volume, head pitch, and hair placement can shift borderline ratios by a few percentage points. Use your primary match and your individual forehead, cheekbone, and jawline bars together when choosing frames or haircuts.
Your selfie is processed locally via WebAssembly and is never uploaded to any server. Verify it yourself: open your browser DevTools (F12), switch to the Network tab, upload a photo, and see that no image request leaves your device.
关于 AI 脸型检测与分析的常见疑问。
