diff --git a/images/example.png b/images/example.png
new file mode 100644
index 0000000..5c18ff2
Binary files /dev/null and b/images/example.png differ
diff --git a/readme.md b/readme.md
index 6a320c4..34ad6b3 100644
--- a/readme.md
+++ b/readme.md
@@ -1,7 +1,38 @@
 ## Accurate 3D Face Reconstruction with Weakly-Supervised Learning: From Single Image to Image Set ##
 
+<p align="center"> 
+<img src="/images/example.png">
+</p>
+
+
 This is an python implement of *Accurate 3D Face Reconstruction with Weakly-Supervised Learning: From Single Image to Image Set*.
 
+The method enforces a hybrid-level weakly-supervised training to achieve accurate CNN-based face reconstruction.
+
+## Features
+
+### Accurate shapes
+The method reconstructs faces with high accuracy. Quantitative evaluations on several benchmarks show its state-of-the-art performance. 
+
+|Method       |FaceWareHouse|Florence|BU3DFE|
+|-------------|-------------|--------|------|
+|[Tewari 17]()|2.19         |-       |-     |
+|[Tewari 18]()|1.84         |-       |-     |
+|[Genova 18]()|-            |1.77    |-     |
+|[Sela 18]()  |-            |-       |2.91  |
+|[PRN]()	  |-            |-       |1.86  |
+|Ours         |1.81         |1.67    |1.40  |
+
+### High fidelity textures
+
+### Robust
+
+### Aligned with images
+
+### Easy and Fast
+Faces are represented with Basel Face Model 2009, which is easy for further manipulations (e.g expression transfer). ResNet-50 is used as backbone network to achieve over 50 fps (on GTX 1080) for reconstructions.
+
+
 ## Getting Started
 ### Prerequisite ###