From 0c70c3c7393bd24793796350e3635b6b04fb95ea Mon Sep 17 00:00:00 2001
From: Anurag Ranjan <anurag.ranjan@tuebingen.mpg.de>
Date: Tue, 8 Nov 2016 00:25:48 +0100
Subject: [PATCH] Updated Readme

---
 README.md | 66 ++++++++++++++++++++++++++++++-------------------------
 1 file changed, 36 insertions(+), 30 deletions(-)

diff --git a/README.md b/README.md
index 9fbd0e0..c2b8aea 100644
--- a/README.md
+++ b/README.md
@@ -1,51 +1,57 @@
 # SPyNet: Spatial Pyramid Network for Optical Flow
-
+This code is based on the paper [Optical Flow Estimation using a Spatial Pyramid Network](https://arxiv.org/abs/1611.00850)
 ## First things first
 You need to have [Torch.](http://torch.ch/docs/getting-started.html#_)
 
 Install other required packages
-
-    cd extras/spybhwd
-    luarocks make
-
-    cd ../stnbhwd
-    luarocks make
-
+```bash
+cd extras/spybhwd
+luarocks make
+cd ../stnbhwd
+luarocks make
+```
 ## For Fast Performace
 #### Set up SPyNet
 Set up SPyNet according to the image size and model. For optimal performance, make sure the image width and height are a multiple of 32. You can also specify your favorite fine tuned model. The present supported modes are `sintelFinal`(default) and `sintelClean`. 
-
-    spynet = require('spynet')
-    computeFlow = spynet.setup(512, 384, 'sintelFinal')
-
+```lua
+spynet = require('spynet')
+computeFlow = spynet.setup(512, 384, 'sintelFinal')
+```
 Now you can call computeFlow anytime to estimate optical flow between image pairs.
 
 #### Computing flow
 Load an image pair and stack it.
-
-    require 'image'
-    im1 = image.load('samples/00001_img1.ppm' )
-    im2 = image.load('samples/00001_img2.ppm' )
-    im = torch.cat(im1, im2, 1)
-
+```lua
+require 'image'
+im1 = image.load('samples/00001_img1.ppm' )
+im2 = image.load('samples/00001_img2.ppm' )
+im = torch.cat(im1, im2, 1)
+```
 SPyNet works with batches of data on CUDA. So, compute flow using
-
-    im = im:resize(1, im:size(1), im:size(2), im:size(3)):cuda()
-    flow = computeFlow(im)
-
+```lua
+im = im:resize(1, im:size(1), im:size(2), im:size(3)):cuda()
+flow = computeFlow(im)
+```
 You can also use batch-mode, if your images `im` are a tensor of size `Bx6xHxW`, of batch size B with 6 RGB pair channels. You can directly use:
-
-    flow = computeFlow(im)
-
+```lua
+flow = computeFlow(im)
+```
 ## For Easy Usage
 ### Coming Soon.
 
 ## Timing Benchmarks
 Our timing benchmark is set up on Flying chair dataset. To test it, you need to download
-
-    wget http://lmb.informatik.uni-freiburg.de/resources/datasets/FlyingChairs/FlyingChairs.zip
-
+```bash
+wget http://lmb.informatik.uni-freiburg.de/resources/datasets/FlyingChairs/FlyingChairs.zip
+```
 Run the timing benchmark
+```bash
+th timing_benchmark.lua -data YOUR_FLYING_CHAIRS_DATA_DIRECTORY
+```
+
+## References
+1. Our warping code is based on [qassemoquab/stnbhwd.](https://github.com/qassemoquab/stnbhwd)
+2. Dosovitskiy, Alexey, et al. "Flownet: Learning optical flow with convolutional networks." 2015 IEEE International Conference on Computer Vision (ICCV). IEEE, 2015.
 
-    th timing_benchmark.lua -data YOUR_FLYING_CHAIRS_DATA_DIRECTORY
-    
+## When using this code, please cite
+Anurag Ranjan and Michael Black, "Optical Flow Estimation using a Spatial Pyramid Network", arXiv preprint arXiv:1611.00850 cs.CV (2016).