poprawa sieci neuronowej oraz drobne zmiany

main.py

@@ -303,7 +303,7 @@ class Game:
                     goal = x//TILE_SIZE,y//TILE_SIZE
                     mob_image = self.sauron.image_path
                     prediction = self.prediction_road(x,y,mob_image)
-                    prediction = "SAURON"
+                    #prediction = "SAURON"
                     while True:   #do poprawienia poprawne rozpoznawanie
                         print("goal: ",goal)
                         if prediction == "SAURON":
@@ -331,7 +331,7 @@ class Game:
                                 prediction = self.prediction_road(x,y,mob_image)
                                 prediction = "ORK_ARCHER"
                         
-                        elif prediction == "ORK_INFANTRY":
+                        elif prediction == "ORK_MELEE":
                             if self.agent.level < self.infantry_ork.level:
                                 lvl = 'nie'
                             else:
@@ -356,7 +356,7 @@ class Game:
                             goal = x//TILE_SIZE,y//TILE_SIZE
                             mob_image = self.sauron.image_path
                             prediction = self.prediction_road(x,y,mob_image)
-                            prediction = "SAURON"
+                            #prediction = "SAURON"
                         elif prediction == "ORK_ARCHER":
                             if self.agent.level < self.archer_ork.level:
                                 lvl = 'nie'
@@ -384,7 +384,7 @@ class Game:
                             goal = x//TILE_SIZE,y//TILE_SIZE
                             mob_image = self.infantry_ork.image_path
                             prediction = self.prediction_road(x,y,mob_image)
-                            prediction = "ORK_INFANTRY"
+                            #prediction = "ORK_INFANTRY"
                          
                         
                            

mobs.py

@@ -14,7 +14,7 @@ class Archer_ork(pygame.sprite.Sprite):
         self.width = TILE_SIZE
         self.height = TILE_SIZE
 
-        self.image_path = "./zdjecia/ORK_ARCHER/ork_lucznik.png"
+        self.image_path = "./zdjecia/ORK_ARCHER/ork_archer (889).jpg"
         self.ARCHER_ORK_IMG = pygame.image.load(self.image_path)
         self.ARCHER_ORK = pygame.transform.scale(self.ARCHER_ORK_IMG,(64,64))  
 
@@ -45,7 +45,7 @@ class Infantry_ork(pygame.sprite.Sprite):
         self.width = TILE_SIZE
         self.height = TILE_SIZE
 
-        self.image_path = "./zdjecia/ORK_MELEE/ork-piechota.png"
+        self.image_path = "C:\\mobs_photos\\ork_melee (11).jpg"#sciezka do zmiany
         self.INFANTRY_ORK_IMG = pygame.image.load(self.image_path)
         self.INFANTRY_ORK = pygame.transform.scale(self.INFANTRY_ORK_IMG,(64,64)) 
 
@@ -77,7 +77,7 @@ class Sauron(pygame.sprite.Sprite):
         self.width = TILE_SIZE
         self.height = TILE_SIZE
 
-        self.image_path = "./zdjecia/SAURON/sauron.png"
+        self.image_path = "C:\\mobs_photos\\sauron (700).jpg"#sciezka do zmiany
         self.SAURON_IMG = pygame.image.load(self.image_path)
         self.SAURON = pygame.transform.scale(self.SAURON_IMG,(64,64))
 

nn.py

@@ -11,8 +11,8 @@ import pathlib
 
 
 class NeuralN:
-#    @staticmethod
-    def predict(self,image_path):
+    # @staticmethod
+    def predict(self, image_path):
         data_dir = pathlib.Path('zdjecia')
         saved_model_path = pathlib.Path('trained_model.h5')
         class_names_path = pathlib.Path("class_names.pkl")
@@ -47,12 +47,6 @@ class NeuralN:
                 image_size=(180, 180),
                 batch_size=32)
 
-            # test_ds = tf.keras.utils.image_dataset_from_directory(
-            #    data_dir,
-            #    seed=123,
-            #    image_size=(180, 180),
-            #    batch_size=32)
-
             class_names = train_ds.class_names
             print(class_names)
 
@@ -77,7 +71,7 @@ class NeuralN:
                           metrics=['accuracy'])
             model.summary()
 
-            epochs = 1
+            epochs = 10
             history = model.fit(
                 train_ds,
                 validation_data=val_ds,
@@ -92,31 +86,24 @@ class NeuralN:
         probability_model = tf.keras.Sequential([model,
                                                  tf.keras.layers.Softmax()])
 
-        #image_path = image
-        image_path = pathlib.Path('zdjecia\ORK_ARCHER\ork_archer (2).png')
+        # image_path = image
+        #image_path = pathlib.Path('')
         image = Image.open(image_path)
 
         # Preprocess the image
+
         image = image.resize((180, 180))  # Resize to match the input size of the model
         image_array = tf.keras.preprocessing.image.img_to_array(image)
         image_array = image_array / 255.0  # Normalize pixel values
 
         # Add an extra dimension to the image array
-        image_array = tf.expand_dims(image_array, 0)
+        image_array = tf.expand_dims(image, 0)
         # Make the prediction
-        predictions = probability_model.predict(image_array)
-
+        model = tf.keras.models.load_model("trained_model.h5")
+        prediction = model.predict(image_array)
         # Convert the predictions to class labels
-        predicted_label = class_names[predictions[0].argmax()]
-        #actions = {
-         #   'ORK_MELEE': 'fight',
-          #  'ORK_ARCHER': 'change_dir',
-           # 'SAURON': 'change_dir'
-        #}
-
-        # Get the action for the predicted character
-        #action = actions.get(predicted_label, 'unknown')
+        predicted_label = class_names[prediction[0].argmax()]
 
         # Print the predicted label
         print(predicted_label)
-        return predicted_label#, action
+        return predicted_label