fix typos

README.md

@@ -4,7 +4,7 @@ This repository contains tensorflow implemenation of two models for semantic seg
 
 * **U-net** (https://arxiv.org/abs/1505.04597) : The network can be found in [u_net.py](./u_net.py). Here is the architecture which I have borrowed from the paper :
 ![unet](./misc/unet.png)
-There are a few minor differences in my implementation. I have used 'same' padding to simplify things. For the upsampling, I have simply used `tf.image.resize_images` function (see [layers_unet.py](./layers_unet.py)) . The full transpose convolution (deconvolution) layer is used for FCN described next. 
+There are a few minor differences in my implementation. I have used 'same' padding to simplify things. For the upsampling, I have simply used `tf.image.resize_images` function (see [layers_unet.py](./layers_unet.py)) . The full transpose convolution (deconvolution) layer is implemented for FCN described next. 
 
 * **FCN** with **global convolution network**  (https://arxiv.org/abs/1703.02719) : The network can be found in [fcn_gcn_net.py](./fcn_gcn_net.py). Here is the architecture which I have borrowed from the paper :
 ![fcn_gcn](./misc/fcn_gcn.png)