Update README.md

Change image position

Author: DylanWusee

README.md

@@ -1,15 +1,15 @@
 # PointPWC-Net: *Cost Volume on Point Clouds for (Self-) Supervised Scene Flow Estimation*
 This is the code for [PointPWC-Net](https://arxiv.org/abs/1911.12408), a deep coarse-to-fine network designed for 3D scene flow estimation from 3D point clouds. Created by <a href="http://web.engr.oregonstate.edu/~wuwen/" target="_blank">Wenxuan Wu</a>, <a href="http://web.engr.oregonstate.edu/~lif/" target="_blank">Fuxin Li</a> from Oregon State University.
 
+<img src="./images/PointPWC.png" width=100%>
+
 **Update:**
 
 * Self-supervised loss is release in model.py(multiScaleChamferSmoothCurvature).
 You can train the self-supervised model by using train_self.py.
 * Update PointConvFlow to compute the patch-to-patch cost volume.
 * Update a updated model pretrain weight to obtain a better result than the original paper.
 
-<img src="./images/PointPWC.png" width=100%>
-
 ## Citation
 
 If you find our work useful in your research, please consider citing: