update

.gitignore

@@ -150,3 +150,4 @@ cython_debug/
 #  and can be added to the global gitignore or merged into this file.  For a more nuclear
 #  option (not recommended) you can uncomment the following to ignore the entire idea folder.
 .idea/
+/algorithms/neural_network/data/

algorithms/neural_network/neural_network.py

@@ -10,23 +10,33 @@ from common.constants import DEVICE, BATCH_SIZE, NUM_EPOCHS, LEARNING_RATE, SETU
 
 class NeuralNetwork(pl.LightningModule):
     def __init__(self, numChannels=3, batch_size=BATCH_SIZE, learning_rate=LEARNING_RATE, num_classes=4):
-        super().__init__()
+        super(NeuralNetwork, self).__init__()
-        self.layer = nn.Sequential(
+        self.conv1 = nn.Conv2d(numChannels, 24, (3, 3), padding=1)
-            nn.Linear(36*36*3, 300),
+        self.relu1 = nn.ReLU()
-            nn.ReLU(),
+        self.maxpool1 = nn.MaxPool2d((2, 2), stride=2)
-            nn.Linear(300, 4),
+        self.conv2 = nn.Conv2d(24, 48, (3, 3), padding=1)
-            nn.LogSoftmax(dim=-1)
+        self.relu2 = nn.ReLU()
-        )
+        self.fc1 = nn.Linear(48*18*18, 4)
+        self.relu3 = nn.ReLU()
+        self.fc2 = nn.Linear(500, num_classes)
+        self.logSoftmax = nn.LogSoftmax(dim=1)
+
         self.batch_size = batch_size
         self.learning_rate = learning_rate
 
     def forward(self, x):
+        x = self.conv1(x)
+        x = self.relu1(x)
+        x = self.maxpool1(x)
+        x = self.conv2(x)
+        x = self.relu2(x)
         x = x.reshape(x.shape[0], -1)
-        x = self.layer(x)
+        x = self.fc1(x)
+        x = self.logSoftmax(x)
         return x
 
     def configure_optimizers(self):
-        optimizer = SGD(self.parameters(), lr=self.learning_rate)
+        optimizer = Adam(self.parameters(), lr=self.learning_rate)
         return optimizer
 
     def training_step(self, batch, batch_idx):

algorithms/neural_network/neural_network_interface.py

@@ -100,7 +100,7 @@ def what_is_it(img_path, show_img=False):
         plt.imshow(plt.imread(img_path))
         plt.show()
     image = SETUP_PHOTOS(image).unsqueeze(0)
-    model = NeuralNetwork.load_from_checkpoint('./lightning_logs/version_3/checkpoints/epoch=8-step=810.ckpt')
+    model = NeuralNetwork.load_from_checkpoint('./lightning_logs/version_13/checkpoints/epoch=4-step=405.ckpt')
 
     with torch.no_grad():
         model.eval()
@@ -109,17 +109,17 @@ def what_is_it(img_path, show_img=False):
 
 
 CNN = NeuralNetwork()
+common.helpers.createCSV()
 
-
+#trainer = pl.Trainer(accelerator='gpu', devices=1, callbacks=[EarlyStopping('val_loss')], max_epochs=NUM_EPOCHS)
-trainer = pl.Trainer(accelerator='gpu', devices=1, auto_scale_batch_size=True, callbacks=[EarlyStopping('val_loss')], max_epochs=NUM_EPOCHS)
+trainer = pl.Trainer(accelerator='gpu', devices=1, auto_lr_find=True, max_epochs=NUM_EPOCHS)
-#trainer = pl.Trainer(accelerator='gpu', devices=1, auto_lr_find=True, max_epochs=NUM_EPOCHS)
 
 trainset = WaterSandTreeGrass('./data/train_csv_file.csv', transform=SETUP_PHOTOS)
 testset = WaterSandTreeGrass('./data/test_csv_file.csv', transform=SETUP_PHOTOS)
 train_loader = DataLoader(trainset, batch_size=BATCH_SIZE, shuffle=True)
 test_loader = DataLoader(testset, batch_size=BATCH_SIZE)
 
-#trainer.fit(CNN, train_loader, test_loader)
+trainer.fit(CNN, train_loader, test_loader)
 #trainer.tune(CNN, train_loader, test_loader)
-check_accuracy_tiles()
+#check_accuracy_tiles()
-print(what_is_it('../../resources/textures/sand.png', True))
+#print(what_is_it('../../resources/textures/sand.png', True))

common/constants.py

@@ -77,19 +77,18 @@ BAR_HEIGHT_MULTIPLIER = 0.1
 
 
 #NEURAL_NETWORK
-LEARNING_RATE = 0.13182567385564073
+LEARNING_RATE = 0.00478630092322638
 BATCH_SIZE = 64
-NUM_EPOCHS = 50
+NUM_EPOCHS = 20
 
 DEVICE = torch.device('cuda') if torch.cuda.is_available() else torch.device('cpu')
 print("Using ", DEVICE)
 CLASSES = ['grass', 'sand', 'tree', 'water']
 
 SETUP_PHOTOS = transforms.Compose([
-    transforms.Resize(36),
-    transforms.CenterCrop(36),
     transforms.ToPILImage(),
     transforms.ToTensor(),
+    transforms.Resize((36, 36)),
     transforms.Normalize([0.5, 0.5, 0.5], [0.5, 0.5, 0.5])
 ])