-
I am unable to understand what are the three outputs of pose. Generally it is [yaw, pitch and roll], but here it is not working.
-
Hi, thank you for your grate work!
Have you trained the model based MS1M-Arcface dataset? I tried to train, only get 0.9755 accuracy on LFW.
Looking forward to your reply, thanks!
-
A Lightweight Face Recognition and Facial Attribute Analysis (Age, Gender, Emotion and Race) Library for Python.It is a hybrid face recognition framework wrapping state-of-the-art models: [VGG-Face](h…
-
Recently, I have used ijb_evals.py to test mobilefacenet based on arcface. The train dataset is CASIA-WebFace and the performance is as follows,and I wonder if the results are reasonable because N1D1F…
-
Hi,
Please check/adopt the following model: https://github.com/deepinsight/insightface/wiki/Model-Zoo
In praticular LResNet100E-IR,ArcFace@ms1m-refine-v2 looks good.
Thanks
-
@nttstar I checked the pre-trained model that was made available and there is no fc7_*_weight layers - referred in Line 100 of drop.py (Using r100's pretrained models)I think these layers comes from p…
-
I tried to calculate the ID retrieval for the code using the pretrained arcface model but got the values very low. I believe my approach to calculate ID retrieval is wrong. I tried with 500 images.
-
hello:
When I was training arcface's second epoch, what was the reason for this?
![image](https://user-images.githubusercontent.com/86965285/165713881-4a7d6932-6a7d-477c-a780-52582954b95a.png)
This…
-
This is a tracking issue for different face embeddings. The main criteria what the cropping procedure is (is it compatible with dlib?) are how well it performs on different benchmarks and the size. LF…
-
PackagesNotFoundError: The following packages are not available from current channels:
- ffmpeg==4.3=hf484d3e_0
- autocrop==1.1.1=pypi_0
- torchvision==0.9.1=py38_cu111
- opencv-python==…