Several problem fixes

.gitignore

@@ -0,0 +1,3 @@
+models
+Params
+__pycache__

Input_mouse_event.py

@@ -73,8 +73,8 @@ def __init__(self, mode_list, paint_size, up_sketch_view , parent=None):
                 self.model[key] = AE_Model()
                 self.model[key].initialize(self.opt)
                 self.model[key].eval()
-
-                self.mask[key] = cv2.cvtColor(cv2.imread('heat/' + key + '.jpg'), cv2.COLOR_RGB2GRAY).astype(np.float) / 255
+                mask_filename = "heat/{0}.jpg".format(key) if (key != '') else "heat/bg.jpg"
+                self.mask[key] = cv2.cvtColor(cv2.imread(mask_filename), cv2.COLOR_RGB2GRAY).astype(np.float) / 255
                 self.mask[key] = np.expand_dims(self.mask[key], axis=2)
 
             #for input and refine weight
@@ -301,4 +301,4 @@ def start_Shadow(self):
 
     def thread_shadow(self):
         while True:
-            self.start_Shadow();
+            self.start_Shadow();

install.sh

@@ -1,6 +1,25 @@
-# install necessary library
-pip install pyqt5
-pip install Pillow 
-pip install scipy 
-pip install dominate 
-pip install opencv-python
+# install necessary library
+pip install pyqt5
+pip install Pillow
+pip install scipy
+pip install dominate
+
+# On many systems Qt for the desktop does not correspond to the
+# libraries that OpenCV was compiled against.  
+#
+# In particular those libraries are
+#
+# libQtDBus
+# libQtCore
+# libQtGui
+#
+# This will cause a segmentation fault when the program is run
+# 
+# see https://github.com/skvark/opencv-python/issues/46
+#
+# In that case, uncomment the following to force pip to build opencv-python
+# from source.
+#
+# pip install --no-binary opencv-python opencv-python
+#
+pip install opencv-python