face detection model failed in the CPU environment

YonghaoHe / LFFD-A-Light-and-Fast-Face-Detector-for-Edge-Devices

A light and fast one class detection framework for edge devices. We provide face detector, head detector, pedestrian detector, vehicle detector......

MIT License

1.31k stars 330 forks source link

face detection model failed in the CPU environment #105

Closed tairenchen closed 3 years ago

tairenchen commented 3 years ago

Thank you for your good work with face detection.

It is working well in the GPU environment but failed in the CPU case.

System: Ubuntu 18.04.5 LTS"

Mxnet: mxnet-mkl==1.6.0

Error information: A lot of bounding boxes are detected on the test image

My changes: In the folder: "LFFD-A-Light-and-Fast-Face-Detector-for-Edge-Devices/face_detection/accuracy_evaluation", I changed the "ctx=mxnet.gpu(0)" to "ctx=mxnet.cpu(0)" in the file: "predict.py"

See attached the image for more information test_result

NgoJunHaoJason commented 3 years ago

@tairenchen does it work if you use another version of mxnet, such as mxnet-cu100mkl==1.5.1?

tairenchen commented 3 years ago

Hi @NgoJunHaoJason

Thank you for your reply.

I used the CPU version, so the version "mxnet-cu100mkl==1.5.1" you mentioned does not work. Because your version works for GPU environment, i.e.: CUDA available.

But I did give it a try, it gives me the error information as below:

tairenchen commented 3 years ago

Okay, the problem is solve when I install

Mxnet: mxnet-mkl==1.5.0

Thank you