diff --git a/.gitignore b/.gitignore
index 6320cd2..ce99277 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1 +1,2 @@
-data
\ No newline at end of file
+data
+__pycache__
\ No newline at end of file
diff --git a/README.md b/README.md
index 0cae1c0..44ed436 100644
--- a/README.md
+++ b/README.md
@@ -10,7 +10,7 @@
- 
+ 
diff --git a/analyze.py b/analyze.py
new file mode 100644
index 0000000..4f77af4
--- /dev/null
+++ b/analyze.py
@@ -0,0 +1,45 @@
+import os
+import torch
+import numpy as np
+import matplotlib.pyplot as plt
+
+from utils.dataloader import val_loader
+from models.medium import mediumNet
+from models.small import smallNet
+
+os.environ["KMP_DUPLICATE_LIB_OK"]="TRUE"
+PATH = './trained_model/medium_74_128px.pth'
+
+# Our Dataset Classes
+classes = ('airplane', 'cat', 'dog', 'motorbike', 'person')
+
+model = mediumNet()
+trained_weight = torch.load(PATH, map_location='cpu')
+model.load_state_dict(trained_weight)
+
+def custom_imshow(imgList, predicted):
+
+ fig = plt.figure()
+
+ rows = 2
+ cols = 2
+
+ for i in range(4):
+ img = imgList[i]
+ temp = fig.add_subplot(rows, cols, i+1)
+ temp.set_title(classes[predicted[i]])
+ temp.imshow(np.transpose(img, (1, 2, 0)))
+ temp.axis('off')
+
+ plt.show()
+
+for idx, data in enumerate(val_loader):
+
+ inputs, labels = data
+ outputs = model(inputs)
+ _, predicted = torch.max(outputs.data, 1)
+
+ custom_imshow(inputs, predicted)
+
+
+
diff --git a/main.py b/main.py
index 467b05e..df6ce41 100644
--- a/main.py
+++ b/main.py
@@ -1,5 +1,5 @@
import torch
-from dataloader import train_loader, val_loader, custom_imshow
+from utils.dataloader import train_loader, val_loader, custom_imshow
import matplotlib.pyplot as plt
import numpy as np
diff --git a/medium.pth b/medium.pth
deleted file mode 100644
index 7dbd243..0000000
Binary files a/medium.pth and /dev/null differ
diff --git a/models/__pycache__/medium.cpython-37.pyc b/models/__pycache__/medium.cpython-37.pyc
index b2fe728..206a2aa 100644
Binary files a/models/__pycache__/medium.cpython-37.pyc and b/models/__pycache__/medium.cpython-37.pyc differ
diff --git a/models/__pycache__/small.cpython-37.pyc b/models/__pycache__/small.cpython-37.pyc
index c4c2a57..0c85755 100644
Binary files a/models/__pycache__/small.cpython-37.pyc and b/models/__pycache__/small.cpython-37.pyc differ
diff --git a/models/medium.py b/models/medium.py
index 49825ce..4651770 100644
--- a/models/medium.py
+++ b/models/medium.py
@@ -15,13 +15,11 @@ def __init__(self) :
self.batchConv3 = nn.BatchNorm2d(16)
self.conv4 = nn.Conv2d(16,32,3)
self.batchConv4 = nn.BatchNorm2d(32)
- self.conv5 = nn.Conv2d(32,48,3)
- self.batchConv5 = nn.BatchNorm2d(48)
self.pool = nn.MaxPool2d(2,2)
- self.fc1 = nn.Linear(768,256)
- self.fc2 = nn.Linear(256,128)
+ self.fc1 = nn.Linear(1152,512)
+ self.fc2 = nn.Linear(512,128)
self.fc3 = nn.Linear(128,5)
self.dropout = nn.Dropout(p=0.5)
@@ -42,7 +40,6 @@ def forward(self, x) :
x = self.pool(F.relu(self.conv2(x)))
x = self.pool(F.relu(self.conv3(x)))
x = self.pool(F.relu(self.conv4(x)))
- x = F.relu(self.conv5(x))
x = x.view(batchSize, -1)
x = self.dropout(x)
diff --git a/models/small.py b/models/small.py
index 6cf5712..b498afd 100644
--- a/models/small.py
+++ b/models/small.py
@@ -14,7 +14,8 @@ def __init__(self) :
self.pool = nn.MaxPool2d(2,2)
self.fc1 = nn.Linear(2304, 1024)
- self.fc2 = nn.Linear(1024,5)
+ self.fc2 = nn.Linear(1024,512)
+ self.fc3 = nn.Linear(512,5)
nn.init.kaiming_normal_(self.fc1.weight)
nn.init.kaiming_normal_(self.fc2.weight)
@@ -29,6 +30,7 @@ def forward(self, x) :
x = x.view(batchSize, -1)
x = F.relu(self.fc1(x))
- x = self.fc2(x)
+ x = F.relu(self.fc2(x))
+ x = self.fc3(x)
return x
\ No newline at end of file
diff --git a/trained_model/small_65_128px.pth b/trained_model/small_65_128px.pth
new file mode 100644
index 0000000..3fc2195
Binary files /dev/null and b/trained_model/small_65_128px.pth differ
diff --git a/dataloader.py b/utils/dataloader.py
similarity index 80%
rename from dataloader.py
rename to utils/dataloader.py
index 0def30c..8c13d5a 100644
--- a/dataloader.py
+++ b/utils/dataloader.py
@@ -29,26 +29,10 @@
]))
val_loader = torch.utils.data.DataLoader(dataset=val_dataset,
- batch_size=10,
- shuffle=False
+ batch_size=4,
+ shuffle=True
)
-def custom_imshow(imgList, labelList):
-
- fig = plt.figure()
-
- rows = 2
- cols = 2
-
- for i in range(4):
- img = imgList[i]
- temp = fig.add_subplot(rows, cols, i+1)
- temp.imshow(np.transpose(img, (1, 2, 0)))
- temp.axis('off')
-
-
- plt.show()
-
if __name__ == "__main__":
for batch_idx, data in enumerate(train_loader) :
inputs, labels = data