diff --git a/datareader_preprocessed.py b/datareader_preprocessed.py
index 7754408..b6c1025 100644
--- a/datareader_preprocessed.py
+++ b/datareader_preprocessed.py
@@ -1,119 +1,120 @@
 import pickle
 import os
 import pickle
 import argparse
 import time
 
 import torch
 from torch.autograd import Variable
 
 import numpy as np
 import project.data_utils as du
 
 # all_files = os.listdir("test/")   # imagine you're one directory above test dir
 # print(all_files)  #
 
 '''
 Reads preprocessed data. For each pedestrian, the coordinates are normalized and the coordinates of pedestrians in same frames are expressed
 relatively to them
 '''
 
 #import utils
 def DataPaths():
     '''
     returns the path of all videos for each scene
     DataPaths[0] = biwi
     DataPaths[1] = crowds
     DataPaths[2] = mot
     DataPaths[3] = stanford
     '''
     #biwi datasets Project/preprocessed_data/biwi/biwi_hotel.txt_5.txt
     biwi_list = os.listdir('preprocessed_data/biwi')
     #print(os.listdir('Project/preprocessed_data/biwi'))
     #print(biwi_list, 'biwi list')
     if 'Icon\r' in biwi_list: #remove the icon file
         biwi_list.remove('Icon\r')
     biwi_files = np.array(biwi_list)  #
     biwi_paths = np.array(['preprocessed_data/biwi/' + x for x in biwi_files])
     #print(biwi_files,'biwi')
     #crowds
     crowds_list = os.listdir('preprocessed_data/crowds')
     if 'Icon\r' in crowds_list: #remove the icon file
         crowds_list.remove('Icon\r')
     crowds_files = np.array(crowds_list)  #
     crowds_paths = np.array(['preprocessed_data/crowds/' + x for x in crowds_files])
 
     #mot
     mot_list = os.listdir('preprocessed_data/mot')
     if 'Icon\r' in mot_list: #remove the icon file
         mot_list.remove('Icon\r')
     mot_files = np.array(mot_list)  #
     mot_paths = np.array(['preprocessed_data/mot/' + x for x in mot_files])
 
     #stanford
     stanford_list = os.listdir('preprocessed_data/stanford')
     if 'Icon\r' in stanford_list: #remove the icon file
         stanford_list.remove('Icon\r')
     stanford_files = np.array(stanford_list)  #
     stanford_paths = np.array(['preprocessed_data/stanford/' + x for x in stanford_files])
 
     path_dataset=[biwi_paths, crowds_paths, mot_paths, stanford_paths]
     return path_dataset
 
 class Pedestrian():
     '''
     Class of pedestrians, instanciated from data text files
 
     Attributes:
     frames: frames where the pedestrian appears
     neighbors : other pedestrians in the frames
     coords : coordinates of the pedestrian in each frame
     neigh_coords: coordinates of the neighbors in each frame
     '''
     def __init__(self, pedID, frames, coords, neighbors, velocity):
         self.pedID = pedID
         self.frames = frames
         self.coords = coords
         self.neighbors = neighbors
         self.velocity = velocity
 
     def getPedID(self):
         return self.pedID
 
     def getFrames(self):
         return self.frames
 
     def getCoords(self):
         return self.coords
 
     def getNeighbors(self):
         return self.neighbors
 
     def getVelocity(self):
         return self.velocity
 
 
 def DataLoader(path_dataset):
     '''
     Returns a list of "Pedestrian" objects for each dataset passed as arg
     '''
 
     datalist = []
     numpeds = len(path_dataset)
 
     for directory in range(0,len(path_dataset)):
         data = np.genfromtxt(path_dataset[directory], dtype = 'float')
         data = torch.from_numpy(data).view(-1,4)
         pedID = data[0,1]
         ped_data= data[(data[:,1]==pedID).nonzero(),:].view(-1,4)
         frames = ped_data[:,0]
         coords = ped_data[:,2:4]
+        print(coords.size())
         neighbors = data[(data[:,1]!=pedID).nonzero(),:].view(-1,4)
         velocity = du.computeVelocity(coords)
-        current_ped = Pedestrian(pedID, frames, coords, neighbors)
+        current_ped = Pedestrian(pedID, frames, coords, neighbors, velocity)
         datalist.append(current_ped)
         #print('Number of frames: %f' %len(frameList))
 
     print('Number of pedestrians: %f' %numpeds)
 
     return datalist