add lfw evalucation and pre-trained model

.gitignore

@@ -99,3 +99,5 @@ ENV/
 
 # mypy
 .mypy_cache/
+
+*.pth

README.md

@@ -1,16 +1,34 @@
 # SphereFace
 A PyTorch Implementation of SphereFace.
-The code can be trained on CASIA-Webface and the best accuracy on LFW is **99.15%**.
+The code can be trained on CASIA-Webface and the best accuracy on LFW is **99.22%**.
 
 [SphereFace: Deep Hypersphere Embedding for Face Recognition](https://arxiv.org/abs/1704.08063)
 
-# Usage
+# Train
 ```
 python train.py
 ```
+
+# Test
+```
+# lfw.tgz to lfw.zip
+tar zxf lfw.tgz; cd lfw; zip -r ../lfw.zip *; cd ..
+
+# lfw evaluation
+python lfw_eval.py --model model/sphere20a_20171020.pth
+```
+
+# Pre-trained models
+| Model name      | LFW accuracy | Training dataset |
+|-----------------|--------------|------------------|
+| [20171020](model/sphere20a_20171020.7z) | 0.9922 | CASIA-WebFace |
+
 # φ
 ![equation](https://latex.codecogs.com/gif.latex?phi%28x%29%3D%5Cleft%28-1%5Cright%29%5Ek%5Ccdot%20%5Ccos%20%5Cleft%28x%5Cright%29-2%5Ccdot%20k)
 
 ![equation](https://latex.codecogs.com/gif.latex?myphi(x)=1-\frac{x^2}{2!}+\frac{x^4}{4!}-\frac{x^6}{6!}+\frac{x^8}{8!}-\frac{x^9}{9!})
 
 ![phi](images/phi.png)
+
+# References
+[sphereface](https://github.com/wy1iu/sphereface)