preparing viper for training: script

ReID/PETA_scripts/viperTraining.py

@@ -0,0 +1,124 @@
+
+# coding: utf-8
+
+# In[2]:
+
+import numpy as np
+import matplotlib.pyplot as plt
+import math
+# get_ipython().magic(u'matplotlib inline')
+import Image
+# Make sure that caffe is on the python path:
+caffe_root = '/users/gpu/agjayant1/caffe-PersonReID/' # Expected to be in rank_scripts/
+
+import sys
+sys.path.insert(0, caffe_root + 'python')
+
+# import caffe
+import sys
+# source_folder = '/home/jayant/vision/PETA/'
+
+import os
+import h5py
+import fileinput
+
+
+# In[3]:
+
+image_data = []
+
+
+# In[4]:
+
+#folder = '/home/jayant/vision/PETA/TownCentre/alex_size/'
+folder = '/users/gpu/agjayant1/PETA/VIPeR/train/'
+
+print "BEGIN"
+Town_images_list = os.listdir(folder)
+
+TC_images_list= []
+for item in Town_images_list:
+    TC_images_list.append((int(item.split('_')[0]),item))
+TC_images_list.sort()
+
+
+for item in TC_images_list:
+    image= item[1]
+    #print image
+    pic = Image.open(folder+image)
+    #A = caffe.io.load_image(folder+image)
+    A = np.array(pic,dtype='f4')
+    A = A.transpose((2,0,1))
+    #B= np.array(A,dtype='f4')
+    image_data.append(A)
+#     image_data.append((image,A))
+
+print "Image Data Loaded"
+# In[5]:
+
+label_data_town=[]
+for line in fileinput.input('/users/gpu/agjayant1/PETA/viper_atr.txt'):
+    label_data_town.append(line[:-2].split(' '))
+
+print "Label Data Loaded"
+# In[6]:
+
+#for item in label_data_GRID:
+#    print item[1:107]
+# check = 6
+# print label_data_town[check]
+# sum1=0
+# for item in label_data_town[check][1:107]:
+#     sum1=sum1+ int(item)
+# print 'sum =',sum1
+# i=1
+# for item in label_data_town[check][1:107]:
+#     if item =='1' :
+#         print i
+#     i=i+1
+
+
+# In[7]:
+
+label = []
+
+
+# In[8]:
+
+for item in TC_images_list:
+    label.append(item[0])
+
+
+# In[9]:
+
+attributes = []
+
+
+# In[10]:
+
+i=0
+for lab in label:
+#     print lab
+    if int(label_data_town[i][0]) == lab:
+            attributes.append(label_data_town[i][1:107])
+    else:
+            i = i+1
+            attributes.append(label_data_town[i][1:107])
+
+
+print "Attributes Loaded"
+# In[11]:
+
+label = np.array(label,dtype='f4')
+attributes = np.array(attributes,dtype='f4')
+
+
+# In[ ]:
+print "Started Writing h5 File "
+h = h5py.File("/users/gpu/agjayant1/PETA/viper_training.h5", 'w' )
+dset1 = h.create_dataset("data", data=image_data)
+dset2 = h.create_dataset("attributes", data=attributes)
+dset3 = h.create_dataset("label",  data=label)
+h.close()
+print "Finished Saving h5 File"
+print "Finishing Script"