From 01a923b26c5d9118e16ac3b25d8f87ea208f156c Mon Sep 17 00:00:00 2001
From: Austin Van Blanton <imaus10@gmail.com>
Date: Fri, 23 Aug 2013 13:27:22 -0400
Subject: [PATCH] Add INRIAPerson dataset

---
 data/INRIAPerson/README.md        |  3 +++
 scripts/downloadDatasets.sh       | 18 ++++++++++++++++++
 scripts/writeINRIAPersonSigset.sh | 40 ++++++++++++++++++++++++++++++++++++++++
 3 files changed, 61 insertions(+), 0 deletions(-)
 create mode 100644 data/INRIAPerson/README.md
 create mode 100755 scripts/writeINRIAPersonSigset.sh

diff --git a/data/INRIAPerson/README.md b/data/INRIAPerson/README.md
new file mode 100644
index 0000000..61d79f6
--- /dev/null
+++ b/data/INRIAPerson/README.md
@@ -0,0 +1,3 @@
+## INRIA Person Database
+Dataset for human detection in several formats: original positive and negative images with bounding box annotations and normalized positive images (just the bounding box).
+* [Website](http://pascal.inrialpes.fr/data/human/)
diff --git a/scripts/downloadDatasets.sh b/scripts/downloadDatasets.sh
index 37daff3..734573c 100755
--- a/scripts/downloadDatasets.sh
+++ b/scripts/downloadDatasets.sh
@@ -35,6 +35,24 @@ if [ ! -d ../data/BioID/img ]; then
   rm *.eye description.txt BioID-FaceDatabase-V1.2.zip
 fi
 
+# INRIA person
+if [ ! -d ../data/INRIAPerson/img ]; then
+  echo "Downloading INRIA person dataset..."
+  if hash curl 2>/dev/null; then
+    curl -OL http://pascal.inrialpes.fr/data/human/INRIAPerson.tar
+  else
+    wget http://pascal.inrialpes.fr/data/human/INRIAPerson.tar
+  fi
+  tar -xf INRIAPerson.tar
+  mkdir ../data/INRIAPerson/img ../data/INRIAPerson/sigset
+  ./writeINRIAPersonSigset.sh Train > ../data/INRIAPerson/sigset/train.xml
+  ./writeINRIAPersonSigset.sh Test > ../data/INRIAPerson/sigset/test.xml
+  ./writeINRIAPersonSigset.sh train_64x128_H96 > ../data/INRIAPerson/sigset/train_normalized.xml
+  ./writeINRIAPersonSigset.sh test_64x128_H96 > ../data/INRIAPerson/sigset/test_normalized.xml
+  mv INRIAPerson/* ../data/INRIAPerson/img
+  rm -r INRIAPerson*
+fi
+
 # KTH
 if [ ! -d ../data/KTH/vid ]; then
   echo "Downloading KTH..."
diff --git a/scripts/writeINRIAPersonSigset.sh b/scripts/writeINRIAPersonSigset.sh
new file mode 100755
index 0000000..cf72936
--- /dev/null
+++ b/scripts/writeINRIAPersonSigset.sh
@@ -0,0 +1,40 @@
+#!/bin/bash
+
+writePresentation()
+{
+    pres="\t\t<presentation Label=\"pos\" file-name=\"$1/pos/$2\""
+    if [ "$#" -eq 6 ]; then
+        pres="$pres>"
+        width=$(($5-$3))
+        height=$(($6-$4))
+        pres="$pres\n\t\t\t<data:bbox height=\"$height\" width=\"$width\" x=\"$3\" y=\"$4\" />"
+        pres="$pres\n\t\t</presentation>"
+    else
+        pres="$pres />"
+    fi
+    printf "$pres\n"
+}
+# for xargs calling bash below
+export -f writePresentation
+SEDREGEX='s/.*(\([0-9]*\), \([0-9]*\)) - (\([0-9]*\), \([0-9]*\))/writeIt \1 \2 \3 \4/'
+
+echo '<?xml version="1.0" encoding="UTF-8"?>'
+echo '<biometric-signature-set>'
+for fullpath in INRIAPerson/$1/pos/*; do
+    filename=$(basename "$fullpath")
+    echo -e "\t<biometric-signature name=\"${filename%.*}\">"
+    if [ -d INRIAPerson/$1/annotations ]; then
+        annotation="INRIAPerson/$1/annotations/${filename%.*}.txt"
+        grep 'Bounding box' $annotation | sed "$SEDREGEX" | xargs -n 5 bash -c "writePresentation $1 $filename \$@"
+    else
+        writePresentation $1 $filename
+    fi
+    echo -e '\t</biometric-signature>'
+done
+for fullpath in INRIAPerson/$1/neg/*; do
+    filename=$(basename "$fullpath")
+    echo -e "\t<biometric-signature name=\"${filename%.*}\">"
+    echo -e "\t\t<presentation Label=\"neg\" file-name=\"$1/neg/$filename\" />"
+    echo -e '\t</biometric-signature>'
+done
+echo '</biometric-signature-set>'
--
libgit2 0.21.4