diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..7e99e36
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1 @@
+*.pyc
\ No newline at end of file
diff --git a/pdm_builder/buildlib/__init__.py b/pdm_builder/buildlib/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/pdm_builder/buildpatch.py b/pdm_builder/buildlib/buildpatch.py
similarity index 99%
rename from pdm_builder/buildpatch.py
rename to pdm_builder/buildlib/buildpatch.py
index db27775..b6ce02e 100644
--- a/pdm_builder/buildpatch.py
+++ b/pdm_builder/buildlib/buildpatch.py
@@ -1,7 +1,7 @@
 from sklearn.svm import SVR, SVC
 from sklearn.linear_model import LogisticRegression
 from PIL import Image
-import config
+import ..config
 import random
 from numpy import array, sqrt, square
 from numpy.linalg import norm
diff --git a/pdm_builder/buildpatch_mosse.py b/pdm_builder/buildlib/buildpatch_mosse.py
similarity index 99%
rename from pdm_builder/buildpatch_mosse.py
rename to pdm_builder/buildlib/buildpatch_mosse.py
index fae86a3..e060426 100644
--- a/pdm_builder/buildpatch_mosse.py
+++ b/pdm_builder/buildlib/buildpatch_mosse.py
@@ -1,5 +1,5 @@
 from PIL import Image
-import config, random
+import ..config, random
 from numpy import array, sqrt, square, fft, mean, zeros
 from numpy.linalg import norm
 from numpy.ma import conjugate
diff --git a/pdm_builder/buildscore.py b/pdm_builder/buildlib/buildscore.py
similarity index 100%
rename from pdm_builder/buildscore.py
rename to pdm_builder/buildlib/buildscore.py
diff --git a/pdm_builder/buildshape.py b/pdm_builder/buildlib/buildshape.py
similarity index 100%
rename from pdm_builder/buildshape.py
rename to pdm_builder/buildlib/buildshape.py
diff --git a/pdm_builder/preprocess.py b/pdm_builder/buildlib/preprocess.py
similarity index 99%
rename from pdm_builder/preprocess.py
rename to pdm_builder/buildlib/preprocess.py
index 6c8a0c1..fa1c7f3 100644
--- a/pdm_builder/preprocess.py
+++ b/pdm_builder/buildlib/preprocess.py
@@ -1,7 +1,7 @@
 import numpy, os
 from numpy import vstack, mean
 from PIL import Image
-import config, procrustes
+import ..config, procrustes
 
 mirror_map = config.mirror_map
 num_patches = config.num_patches
diff --git a/pdm_builder/procrustes.py b/pdm_builder/buildlib/procrustes.py
similarity index 100%
rename from pdm_builder/procrustes.py
rename to pdm_builder/buildlib/procrustes.py
diff --git a/pdm_builder/config.py b/pdm_builder/config.py
index f1b1d5c..fc5eb73 100644
--- a/pdm_builder/config.py
+++ b/pdm_builder/config.py
@@ -1,9 +1,4 @@
 # mirror scheme
-#mirror_map = [14,13,12,11,10,9,8,7,6,5,4,3,2,1,0,19,20,21,22,15,16,17,18,\
-#    28,29,30,31,32,23,24,25,26,27,41,40,39,38,37,36,35,34,33,43,42,50,49,\
-#    48,47,46,45,44,55,54,53,52,51,58,57,56,61,60,59,62,67,68,69,70,63,64,\
-#    65,66]
-    
 mirror_map = [14,13,12,11,10,9,8,7,6,5,4,3,2,1,0,19,20,21,22,15,16,17,18,\
     28,29,30,31,32,23,24,25,26,27,33,40,39,38,37,36,35,34,41,43,42,50,49,\
     48,47,46,45,44,55,54,53,52,51,58,57,56,61,60,59,62,67,68,69,70,63,64,\
@@ -17,11 +12,9 @@
       [19,20,21,22],\
       [23,63,24,64,25,65,26,66,23],\
       [28,67,29,68,30,69,31,70,28],\
-      #[33,34,35,36,42,37,43,38,39,40,41],\
       [34,35,36,42,37,43,38,39,40],\
       [33, 41, 62],\
       [44,45,46,47,48,49,50,51,52,53,54,55,44,56,57,58,50,59,60,61,44],\
-      #62,27,32\
       27,32\
    ]   \
 }
@@ -40,8 +33,9 @@
 #patch_size = 16
 
 # raw image folder
-data_folder = "./newdata/"
-#data_folder = "./data/"
+data_folder = "./data/"
+annotations = "./data/annotations.csv"
+
 # folder image
 cropped_image_folder = "./cropped/"
 
diff --git a/pdm_builder/average_face/average.py b/pdm_builder/create_average_face.py
similarity index 89%
rename from pdm_builder/average_face/average.py
rename to pdm_builder/create_average_face.py
index 8640750..9d3fae0 100644
--- a/pdm_builder/average_face/average.py
+++ b/pdm_builder/create_average_face.py
@@ -1,11 +1,11 @@
 import numpy as np
-import preprocess, pickle, skimage
+from buildlib import preprocess
+import pickle, skimage
 from skimage.transform import PiecewiseAffineTransform, warp
 from skimage.io import imread, Image, imsave
+import config
 
-coordinate_file = "./data.csv"
-
-data_pca, data_patches, meanshape, cropsize = preprocess.preprocess(coordinate_file, mirror = True)
+data_pca, data_patches, meanshape, cropsize = preprocess.preprocess(config.annotations, mirror = True)
 
 dp = {'data_pca' : data_pca, 'data_patches' : data_patches, 'meanshape' : meanshape, 'cropsize' : cropsize}
 fi = open("out.data", "w")
@@ -48,7 +48,7 @@
 avim /= imlen
 avim *= 255  
 avim = avim.astype(np.uint8)
-imsave("./averageface/average.bmp", Image(avim))
+imsave("./average.bmp", Image(avim))
 # for all in array
   # average
   
diff --git a/pdm_builder/emotion_classifier/buildEmotionClass.py b/pdm_builder/create_emotion_classifier.py
similarity index 97%
rename from pdm_builder/emotion_classifier/buildEmotionClass.py
rename to pdm_builder/create_emotion_classifier.py
index 46f0b13..507fee1 100644
--- a/pdm_builder/emotion_classifier/buildEmotionClass.py
+++ b/pdm_builder/create_emotion_classifier.py
@@ -1,9 +1,9 @@
-import preprocess, os
+from buildlib import preprocess
 from sklearn.decomposition import PCA
 from sklearn.linear_model import LogisticRegression
 from sklearn.ensemble import RandomForestClassifier
 from sklearn import cross_validation
-import pickle, json
+import pickle, json, os
 from numpy import row_stack, any, isnan, sum, diag, cov, mean, dot, trace
 from numpy.linalg import qr
 import numpy, math
diff --git a/data/annotations.csv b/pdm_builder/data/annotations.csv
similarity index 100%
rename from data/annotations.csv
rename to pdm_builder/data/annotations.csv
diff --git a/data/readme.txt b/pdm_builder/data/readme.txt
similarity index 60%
rename from data/readme.txt
rename to pdm_builder/data/readme.txt
index e68808e..5eb4c50 100644
--- a/data/readme.txt
+++ b/pdm_builder/data/readme.txt
@@ -1,2 +1,3 @@
 training images for the annotations can be downloaded from https://code.google.com/p/muct/downloads/list
-download muct-e-jpg-v1.tar.gz, muct-c-jpg-v1.tar.gz, muct-b-jpg-v1.tar.gz and muct-a-jpg-v1.tar.gz
\ No newline at end of file
+download muct-e-jpg-v1.tar.gz, muct-c-jpg-v1.tar.gz, muct-b-jpg-v1.tar.gz and muct-a-jpg-v1.tar.gz
+unzip the files and put all training images into the folder "/data/images"
\ No newline at end of file
diff --git a/pdm_builder/main.py b/pdm_builder/pdm_builder.py
similarity index 72%
rename from pdm_builder/main.py
rename to pdm_builder/pdm_builder.py
index a99c3bc..3ed8da3 100644
--- a/pdm_builder/main.py
+++ b/pdm_builder/pdm_builder.py
@@ -1,35 +1,31 @@
-import preprocess
+from buildlib import preprocess, buildshape
 import config
-import buildshape
-#from buildpatch_mosse import build_patches
-from buildpatch import build_patches
+#from buildlab.buildpatch_mosse import build_patches
+from buildlab.buildpatch import build_patches
+from buildlab.buildscore import getScoring
 import pickle
 import numpy, json
-from buildscore import getScoring
 
 buildPatches = True
 buildScoring = True
 
-coordinate_file = "./data.csv"
-
 # preprocess data
-#data_pca, data_patches, meanshape, cropsize = preprocess.preprocess(coordinate_file, mirror = True)
+data_pca, data_patches, meanshape, cropsize = preprocess.preprocess(config.annotations, mirror = True)
 
 #dp = {'data_pca' : data_pca, 'data_patches' : data_patches, 'meanshape' : meanshape, 'cropsize' : cropsize}
 #fi = open("out.data", "w")
 #pickle.dump(dp, fi)
 #fi.close()
 
-fi = open("out.data", "r")
-data = pickle.load(fi)
-fi.close()
-data_pca = data['data_pca']
-data_patches = data['data_patches']
-meanshape = data['meanshape']
-cropsize = data['cropsize']
+#fi = open("out.data", "r")
+#data = pickle.load(fi)
+#fi.close()
+#data_pca = data['data_pca']
+#data_patches = data['data_patches']
+#meanshape = data['meanshape']
+#cropsize = data['cropsize']
 
 # build shape model
-#eigenVectors, eigenValues = buildshape.pca(data_pca.values(), num_components=7)
 eigenVectors, eigenValues = buildshape.pca(data_pca.values(), num_components=24)
 #eigenVectors, eigenValues = buildshape.spca(data_pca.values(), num_components=10, alpha=0.5)
 mean = [numpy.mean(column) for column in meanshape.T]
@@ -59,11 +55,8 @@
 	model['scoring'] = scoring
 model['shapeModel'] = {}
 model['shapeModel']['eigenVectors'] = eigenVectors.T.tolist()
-#model['shapeModel']['eigenValues'] = eigenValues.flatten().tolist()
 model['shapeModel']['eigenValues'] = eigenValues
-#model['shapeModel']['eigenValues'] = eigenValues.tolist()
 model['shapeModel']['meanShape'] = meanshape.tolist()
-#model['shapeModel']['numEvalues'] = eigenValues.shape[0]
 model['shapeModel']['numEvalues'] = len(eigenValues)
 model['shapeModel']['numPtsPerSample'] = meanshape.shape[0]
 
@@ -78,7 +71,7 @@
 
 model['path'] = config.path
 
-of = open("filter.js","w")
+of = open("model.js","w")
 of.write("var pModel = ")
 of.write(json.dumps(model, indent=2))
 of.write(";\n")
diff --git a/readme.md b/readme.md
index e69de29..8f72ec9 100644
--- a/readme.md
+++ b/readme.md
@@ -0,0 +1,7 @@
+
+Dependencies for pdmbuilder is:
+* numpy
+* scipy
+* scikit-learn
+* scikit-image
+* PIL
\ No newline at end of file