Update network model structure:

Changed model from FCNN to CNN

AI/neural_network.py

@@ -1,18 +1,20 @@
-import PIL
-import numpy as np
 import torch
-import torch.nn as nn
-import torch.optim as optim
 import torchvision
 import torchvision.transforms as transforms
+import torch.nn as nn
+import torch.nn.functional as F
+import torch.optim as optim
+import numpy as np
+from matplotlib.pyplot import imshow
+import os
+import PIL
+import numpy as np
 from matplotlib.pyplot import imshow
-
 
 def to_negative(img):
     img = PIL.ImageOps.invert(img)
     return img
 
-
 class Negative(object):
     def __init__(self):
         pass
@@ -20,41 +22,46 @@ class Negative(object):
     def __call__(self, img):
         return to_negative(img)
        
-
-def plotdigit(image):
-    img = np.reshape(image, (-1, 100))
-    imshow(img, cmap='Greys')
-
-
 transform = transforms.Compose([Negative(), transforms.ToTensor()])
-train_set = torchvision.datasets.ImageFolder(root='../src/train', transform=transform)
-classes = ("apple", "potato")
+train_set = torchvision.datasets.ImageFolder(root='train', transform=transform)
+classes = ("pepper", "potato", "strawberry", "tomato")
 
-BATCH_SIZE = 2
+BATCH_SIZE = 4
 train_loader = torch.utils.data.DataLoader(train_set, batch_size=BATCH_SIZE, shuffle=True, num_workers=0)
 
 device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
 
-
 class Net(nn.Module):
     def __init__(self):
-        super(Net, self).__init__()
-        self.flatten = nn.Flatten()
-        self.linear_relu_stack = nn.Sequential(
-            nn.Linear(3 * 100 * 100, 512),
+        super().__init__()
+        self.network = nn.Sequential(
+            nn.Conv2d(3, 32, kernel_size=3, padding=1), #3 channels to 32 channels
             nn.ReLU(),
-            nn.Linear(512, 512),
+            nn.Conv2d(32, 64, kernel_size=3, stride=1, padding=1),
             nn.ReLU(),
-            nn.Linear(512, 2),
-            nn.ReLU()
-        )
-        self.linear_relu_stack = self.linear_relu_stack.to(device)
+            nn.MaxPool2d(2, 2), # output: 64 channels x 50 x 50 image size - decrease
 
-    def forward(self, x):
-        x = self.flatten(x).to(device)
-        logits = self.linear_relu_stack(x).to(device)
-        return logits
+            nn.Conv2d(64, 128, kernel_size=3, stride=1, padding=1),
+            nn.ReLU(),
+            nn.Conv2d(128, 128, kernel_size=3, stride=1, padding=1), #increase power of model 
+            nn.ReLU(),
+            nn.MaxPool2d(2, 2), # output: 128 x 25 x 25
 
+            nn.Conv2d(128, 256, kernel_size=3, stride=1, padding=1),
+            nn.ReLU(),
+            nn.Conv2d(256, 256, kernel_size=3, stride=1, padding=1),
+            nn.ReLU(),
+            nn.MaxPool2d(5, 5), # output: 256 x 5 x 5
+
+            nn.Flatten(), #a single vector 256*5*5,
+            nn.Linear(256*5*5, 1024),
+            nn.ReLU(),
+            nn.Linear(1024, 512),
+            nn.ReLU(),
+            nn.Linear(512, 4))
+        
+    def forward(self, xb):
+        return self.network(xb)
  
 def training_network():
     net = Net()
@@ -63,7 +70,7 @@ def training_network():
     criterion = nn.CrossEntropyLoss()
     optimizer = optim.SGD(net.parameters(), lr=0.001, momentum=0.9)
  
-    for epoch in range(4):
+    for epoch in range(10):
         running_loss = 0.0
         for i, data in enumerate(train_loader, 0):
             inputs, labels = data[0].to(device), data[1].to(device)
@@ -74,7 +81,7 @@ def training_network():
             optimizer.step()
  
             running_loss += loss.item()
-            if i % 2000 == 1999:
+            if i % 200 == 199:
                 print('[%d, %5d] loss: %.3f' % (epoch + 1, i + 1, running_loss))
                 running_loss = 0.0
  
@@ -84,8 +91,8 @@ def training_network():
  
 def result_from_network(net, loaded_image):
     image = PIL.Image.open(loaded_image)
-    pil_to_tensor = transforms.ToTensor()(image.convert("RGB")).unsqueeze_(0)
-    outputs = net(pil_to_tensor.to(device))
+    pil_to_tensor = transforms.Compose([Negative(), transforms.ToTensor()])(image.convert("RGB")).unsqueeze_(0)
+    outputs = net(pil_to_tensor)
  
     return classes[torch.max(outputs, 1)[1]]
  
@@ -99,7 +106,6 @@ def load_network_from_structure(network):
     network.load_state_dict(torch.load('network_model.pth'))
  
  
-# Create network_model.pth
 if __name__ == "__main__":
     print(torch.cuda.is_available())
     training_network()