于老师，您好，求解scale及min_neighbors具体含义？Demo中默认参数有误识别。

[zhehan54]

原图：

[ShiqiYu]

scale：每次缩小图像的比例，不建议修改 min_neighbors：neighbors是检测出的人脸框属性，越大表示是人脸可能性越大。小于min_neighbors的人脸框将被过滤掉。

On Tue, Dec 12, 2017 at 10:50 AM, zhehan54 notifications@github.com wrote:

[image: image] https://user-images.githubusercontent.com/9295204/33864896-e7b63cec-df29-11e7-9467-fa24ffa375de.png [image: image] https://user-images.githubusercontent.com/9295204/33864897-e7c9f098-df29-11e7-846b-a7ad29c4253a.png

原图： [image: web] https://user-images.githubusercontent.com/9295204/33864936-20752c5a-df2a-11e7-8253-671a70525068.jpg

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/ShiqiYu/libfacedetection/issues/50, or mute the thread https://github.com/notifications/unsubscribe-auth/AG0eHRXeih3DY3q-a6L0hv8GkQ2ipVwvks5s_encgaJpZM4Q-V74 .

-- Dr. Shiqi YU (于仕琪) College of Computer Science and Software Engineering, Shenzhen University, 518060, China +86-755-2673 3127

[zhehan54]

@ShiqiYu 谢谢于老师 scale 缩放的原始图像大小是 函数调用时传的参数？就是Demo中的48？倒数第三个参数？ pResults = facedetect_multiview_reinforce(pBuffer, (unsigned char*)(gray.ptr(0)), gray.cols, gray.rows, (int)gray.step, 1.2f, 2, 48, 0, doLandmark);

[sdlyfjx]

scale是缩放大小，就是那个1.2f吧。每次框的大小都乘以1.2。你说的48是最小人脸框的阈值