From d006da427bb6d8970a8d463a7385f4d2f315783c Mon Sep 17 00:00:00 2001
From: Jure Zbontar <jure.zbontar@gmail.com>
Date: Fri, 30 Oct 2015 11:40:10 +0100
Subject: [PATCH] predict_kitti script

---
 README.md         |  6 +++++
 predict_kitti.lua | 64 +++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 70 insertions(+)
 create mode 100755 predict_kitti.lua

diff --git a/README.md b/README.md
index 204f099..80a3d9e 100644
--- a/README.md
+++ b/README.md
@@ -97,6 +97,12 @@ The resulting disparity maps should look like this:
 - `right.png`  
   <img src="samples/output/right_cnn.png" style="width: 25%;"/>
 
+Note that `-disp_max 70` is used only as an example. To reproduce our
+results on the KITTI data sets use `-disp_max 228`.
+
+See the [predict_kitti.lua](predict_kitti.lua) script for how you might
+call `main.lua` in a loop, for multiple image pairs.
+
 ### Load the Output Binary Files ###
 
 You can load the binary files (if, for example, you want to apply 
diff --git a/predict_kitti.lua b/predict_kitti.lua
new file mode 100755
index 0000000..62b7411
--- /dev/null
+++ b/predict_kitti.lua
@@ -0,0 +1,64 @@
+#! /usr/bin/env luajit
+
+--[[
+
+This script computes the 3 pixel error on all KITTI 2012 training examples with
+the fast architecture.
+
+Don't use this script to fit hyperparameters; the error is computed on the
+training examples.
+
+This is not the fastest way to use the neural network---a new process is
+spawned and the network is loaded from disk for each image pair---but
+is probably the safest.
+
+Usage
+-----
+
+   $ ./predict_kitti.lua
+   0   0.0028267929719645
+   1   0.026568045683624
+   2   0.039333925127797
+   ...
+   191 0.078452818068974
+   192 0.012351983422143
+   193 0.066736774940625
+   0.03222369495401
+
+]]--
+
+require 'image'
+require 'torch'
+require 'libadcensus'
+
+path = 'data.kitti/unzip'
+cmd = './main.lua kitti fast -a predict' ..
+   ' -net_fname net/net_kitti_fast_-a_train_all.t7' ..
+   ' -left %s -right %s -disp_max 228'
+
+err_sum = 0
+n_te = 194
+for i = 0, n_te - 1 do
+   -- call mc-cnn
+   local im0 = ('%s/training/image_0/%06d_10.png'):format(path, i)
+   local im1 = ('%s/training/image_1/%06d_10.png'):format(path, i)
+   local im = image.loadPNG(im0)
+   local img_height = im:size(2)
+   local img_width = im:size(3)
+   os.execute(cmd:format(im0, im1) .. ' > /dev/null')
+   local disp = torch.FloatTensor(torch.FloatStorage('disp.bin')):view(1, 1, img_height, img_width)
+
+   -- ground truth
+   local ground_truth = torch.FloatTensor(1, img_height, img_width)
+   adcensus.readPNG16(ground_truth, ('%s/training/disp_noc/%06d_10.png'):format(path, i))
+
+   -- compute the error
+   local mask = torch.ne(ground_truth, 0):float()
+   local bad = torch.add(disp, -1, ground_truth):abs():gt(3):float():cmul(mask)
+   local err = bad:sum() / mask:sum()
+   err_sum = err_sum + err
+   print(i, err)
+
+   collectgarbage()
+end
+print(err_sum / n_te)