FaceQuality

FaceQuality copied to clipboard

Is there any pretrained model that we can download (not baidu)

MyraBaba

In the paper, you said that it would be decayed by 10 after 30, 60, 90 epoch for total of 100 epochs. But in the code, I saw that you...

v-trungdt21

作者您好，如题。希望能提供代码，方便复现工作，蟹蟹～

kongzijian

https://github.com/deepinsight/insightface/blob/master/recognition/common/rec2image.py

Abdurrahheem

about Implementation

10

Hello,I am very interested in your work.and I want to ask some questions, why is the implementation of the code inconsistent with the description in the paper, such as training...

WWWJL

数据集转换

1

你好，我在你给的连接里下载了faces_webface_112x112数据集。但是你给的rec2image.py连接没有了。现在不能生成训练文件。您能再提供一下rec2image.py吗？

ChengBo5

Thank you for perfect repository

huytuong010101

I see the config in step 2 is: ... BACKBONE_RESUME_ROOT = './backbone_resume.pth' HEAD_RESUME_ROOT = './head_resume.pth' TRAIN_FILES = './dataset/face_train_ms1mv2.txt' BACKBONE_LR = 0.05 PRETRAINED_BACKBONE = '' PRETRAINED_QUALITY = '' ... So where...

huytuong010101

Onnx Converted model

2

Hi, I try to conver to onnx but not succesfull ? the bacbone model only results 512 but it should more Best

MyraBaba

Hi, Thanks for your amazing work. However, I have a question why does this implementation use focal loss instead of the loss function below that we see in the paper?

Maisiechiu