From ebdc542442d42ddf99a547b75fee0c8d4ebcfdcc Mon Sep 17 00:00:00 2001
From: Jia Guo <guojia@gmail.com>
Date: Wed, 24 Jan 2018 17:03:48 +0800
Subject: [PATCH 1/4] Update README.md

---
 README.md | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/README.md b/README.md
index 6ee2275..601e705 100644
--- a/README.md
+++ b/README.md
@@ -61,7 +61,8 @@
 ```
 
 ### Requirements
-      1. The only requirement is `MXNet` with GPU support(Python 2.7).
+      1. Install `MXNet` with GPU support(Python 2.7).
+      2. If you want to align dataset by yourself, install tensorflow as we're using the MTCNN writen in TF.
 
 ### Installation
    1. Install MXNet by 

From cc7f42b392abbdfb2d6e3873a4b044451c2740f5 Mon Sep 17 00:00:00 2001
From: Jia Guo <guojia@gmail.com>
Date: Wed, 24 Jan 2018 17:04:38 +0800
Subject: [PATCH 2/4] Update README.md

---
 README.md | 1 +
 1 file changed, 1 insertion(+)

diff --git a/README.md b/README.md
index 601e705..09fa6c0 100644
--- a/README.md
+++ b/README.md
@@ -62,6 +62,7 @@
 
 ### Requirements
       1. Install `MXNet` with GPU support(Python 2.7).
+      
       2. If you want to align dataset by yourself, install tensorflow as we're using the MTCNN writen in TF.
 
 ### Installation

From 8d183dfb2f7f34c7e8a110605beaa3d09deb5fd2 Mon Sep 17 00:00:00 2001
From: Jia Guo <guojia@gmail.com>
Date: Wed, 24 Jan 2018 18:57:05 +0800
Subject: [PATCH 3/4] Update README.md

---
 README.md | 25 +++++++++++++++----------
 1 file changed, 15 insertions(+), 10 deletions(-)

diff --git a/README.md b/README.md
index 09fa6c0..b3b1acf 100644
--- a/README.md
+++ b/README.md
@@ -31,15 +31,19 @@
 
    In InsightFace, we support several popular network backbones and can be set just in one parameter. Below is the list until today:
 
-- ResNet
+* ResNet
+* MobiletNet
+* InceptionResNetV2
+* DPN
+* DenseNet
 
-- MobiletNet
-
-- InceptionResNetV2
-
-- DPN
-
-- DenseNet
+ We also support most of popular face recognition algorithms(losses), by specifying loss type:
+ 
+ * loss-type=0:  Softmax
+ * loss-type=1:  SphereFace
+ * loss-type=2:  AMSoftmax/CosineFace
+ * loss-type=4:  Ours(InsightFace)
+ * loss-type=12: TripletLoss
 
  In our paper, we found there're overlap identities between facescrub dataset and Megaface distractors which greatly affects the identification performance. Sometimes more than 10 percent improvement can be achieved after removing these overlaps. This list will be made public soon in this repository.
 
@@ -61,9 +65,10 @@
 ```
 
 ### Requirements
-      1. Install `MXNet` with GPU support(Python 2.7).
+      
+   1. Install `MXNet` with GPU support(Python 2.7).
       
-      2. If you want to align dataset by yourself, install tensorflow as we're using the MTCNN writen in TF.
+   2. If you want to align dataset by yourself, install tensorflow as we're using the tf-version MTCNN. (Note that any public available MTCNN can be used to align the faces and then transform to 112x112 crop, performance/result should not change.)
 
 ### Installation
    1. Install MXNet by 

From 86e105231fac1b6e5b30e2c8c2b29d8cb22066b9 Mon Sep 17 00:00:00 2001
From: Jia Guo <guojia@gmail.com>
Date: Wed, 24 Jan 2018 20:50:43 +0800
Subject: [PATCH 4/4] Update README.md

---
 README.md | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/README.md b/README.md
index b3b1acf..e568f6d 100644
--- a/README.md
+++ b/README.md
@@ -64,6 +64,17 @@
 }
 ```
 
+  If you want to download the refined MS1M dataset we provided, please cite the paper below:
+  
+```
+@INPROCEEDINGS { guo2016msceleb,
+            author = {Guo, Yandong and Zhang, Lei and Hu, Yuxiao and He, Xiaodong and Gao, Jianfeng},
+            title = {M{S}-{C}eleb-1{M}: A Dataset and Benchmark for Large Scale Face Recognition},
+            booktitle = {European Conference on Computer Vision},
+            year = {2016},
+            organization={Springer}}
+```
+
 ### Requirements
       
    1. Install `MXNet` with GPU support(Python 2.7).