Face-Detector-1MB-with-landmark
Face-Detector-1MB-with-landmark copied to clipboard
biubug6
mobilenet0.25 c++ inference problem
there is no face detected when I change the model to mobilenet0.25 and the nodes name of out out1 out2 to output0 586 585.
when I change the model to mobilenet0.25 and the nodes name of out out1 out2 to output0 586 585.there are many wrong face detected. Have you solved this problem
change https://github.com/biubug6/Face-Detector-1MB-with-landmark/blob/master/Face_Detector_ncnn/FaceDetector.cpp#L45 false, there are also many wrong face detected,but the number is different from above.
我本地改动run_video_face_detect_onnx.py 文件支持测试图片，使用Mb_Tiny_RFB_FD_train_input_640.onnx模型，但是结果未检测到人脸，这个是为什么呢？@biubug6
@Royzon @DuckJ Thank you for your feedback. The anchor size of retinaface is different from RFB or Slim. I will fix bug today.
@Royzon @DuckJ Bug has been repaired! If you want to use retinaface C++ inference, please use "Detector detector(param, bin, true);" in main.cpp.
@biubug6 thx for your greate works @DuckJ For mobilenet0.25_Final.pth c++ inference, my steps:
- change the model to mobilenet0.25 and the nodes name of out out1 out2 to output0 586 585
- use "Detector detector(param, bin, true);" in main.cpp. and I set max_side=640(keep same as the training step??)
- FaceDetector.cpp -->create_anchor_retinaface (keep same as the training step) L212 minsize1={16,32}; L214 minsize1={64,128}; L216 minsize1={256,512};
作者原模型的landmark输出只有4个点，为什么却说是5个点的关键点呢？--------在face.param里面找到作者对应的几个输出口，就可以看到输出维度！
