MTCNN

MTCNN copied to clipboard

Could you explain the parameters of those functions? I' am so confused when trying to understand these algorithms :) Should I examine the model output layer? Could you give me...

yasintuncerr

When I run **make** in **build** , it happens "warning: libopencv_core.so.3.2, needed by /data/caffe_wo/lib/libcaffe.so, may conflict with libopencv_core.so.2.4 /usr/bin/ld: CMakeFiles/mtcnn_c.dir/main.cpp.o: undefined reference to symbol '_ZN2cv6String10deallocateEv' //usr/local/lib/libopencv_core.so.3.2: error adding symbols: DSO...

flexibility2

Does V2 work in current implementation. I am getting segmentation fault.

manish-grafty

Can I please know if the code does face alignment also after face detection. If so can I please know the line number in main.cpp where face alignment code starts

ashokbugude

在识别多人的图片时使用V2 Model会有异常

1

用下面的两张图。在识别多人的图片时使用V2 Model会有异常，根踪到这里应是矩型数据出现了问题， ``` cv::Mat tmp(rect.height, rect.width, CV_32F, cv::Scalar(0.0)); sample_norm_channels[k](cuted_rect).copyTo(tmp(inner_rect)); cv::resize(tmp, n_channels[i * 5 * num_channels_ + k * num_channels_ + j], cv::Size(24, 24)); ```   ]()

wyd1520

How to use v2? Just changing model_v1 to model_v2 gives me segmentation fault.

2

yxchng

jiagh2010

Is the source code support cuda?

1

anguoyang