diff --git a/evaluate.py b/evaluate.py
new file mode 100644
index 00000000..bb2f3e8d
--- /dev/null
+++ b/evaluate.py
@@ -0,0 +1,64 @@
+import h5py
+import scipy.io as sio
+import PIL.Image as Image
+import numpy as np
+import os
+import glob
+import torchvision.transforms.functional as F
+from image import *
+from model import CSRNet
+import torch
+import time
+
+
+from torchvision import datasets, transforms
+transform=transforms.Compose([transforms.ToTensor(),
+    transforms.Normalize(mean=[0.485, 0.456, 0.406],std=[0.229, 0.224, 0.225])])
+
+
+data_path='models/shanghaiTech_Crowd_Counting_Dataset/part_A_final/test_data/images/'
+img_paths = glob.glob(os.path.join(data_path, '*.jpg'))
+
+
+if torch.cuda.is_available():
+    use_cuda = True
+else:
+    use_cuda = False
+
+
+model = CSRNet(load_weights=True)
+
+if use_cuda:
+    model = model.cuda()
+    checkpoint = torch.load('models/partBmodel_best.pth.tar')
+else:
+    checkpoint = torch.load('models/partBmodel_best.pth.tar', map_location='cpu')
+model.load_state_dict(checkpoint['state_dict'])
+
+
+mae = 0
+for i in xrange(len(img_paths)):
+    t1 = time.time()
+    
+    if use_cuda:
+        img = transform(Image.open(img_paths[i]).convert('RGB')).cuda()
+    else:
+        img = transform(Image.open(img_paths[i]).convert('RGB'))
+
+    file_name = img_paths[i].replace('.jpg','.mat').replace('images','ground_truth')
+    gt_file_name = os.path.join(os.path.dirname(file_name), 'GT_' + os.path.basename(file_name))
+    
+    gt_file = sio.loadmat(gt_file_name)
+    #print(gt_file)
+    #print(len(gt_file['image_info'][0][0][0][0][0]))
+    #print(gt_file['image_info'][0][0][0][0][1][0][0])
+    groundtruth = gt_file['image_info'][0][0][0][0][1][0][0]
+
+    output = model(img.unsqueeze(0))
+    t2 = time.time()
+    print(t2-t1)
+    print('{}  --  {} : {}'.format(img_paths[i], output.detach().cpu().sum().numpy(), groundtruth))
+    mae += abs(output.detach().cpu().sum().numpy()-np.sum(groundtruth))
+    #print i,mae
+print mae/len(img_paths)
+