search

biubug6 / Pytorch_Retinaface

Retinaface get 80.99% in widerface hard val using mobilenet0.25.
MIT License
2.63k stars 774 forks source link

if image resolution is very big, detect results is very bad? #183

Open Dian-Yi opened 2 years ago

Dian-Yi commented 2 years ago

I use detect.py to detect my test image, but the result is bad. i use resnet50 Do I need to manually set the resolution of the network input? Test Image: 72af89ba6022a8650e4764bcbe2512014af8cbf5ec7263437757d512

Result: test

huashunfu commented 2 years ago

我请问一下，您直接使用python setyup.py build_ext --inplace生成.so文件后就能import bbox_overlaps吗

huashunfu commented 2 years ago

一直没调起来，如果您知道的话麻烦告诉一下，谢谢了~！

Dian-Yi commented 2 years ago

一直没调起来，如果您知道的话麻烦告诉一下，谢谢了~！

我请问一下，您直接使用python setyup.py build_ext --inplace生成.so文件后就能import bbox_overlaps吗

没有 跑python setyup.py build_ext --inplace 这个，直接搭好环境跑脚本就行了

JavanehBahrami commented 11 months ago

Hi I have the same problem for face images which are close to the camera the face detection results are not good. I run this coommand:

python3 detect.py --trained_model ./weights/Resnet50_Final.pth --network resnet50  --save_image

what do you mean by using python setyup.py build_ext --inplace? most of my images are faces that are close to the camera, What can I do? my test image size (3456, 4608) @Dian-Yi @huashunfu @xsacha @benuri

1