src/import torch.py

@@ -1,3 +0,0 @@
-import torch
-x = torch.rand(5, 3)
-print(x)

src/main.py

@@ -1,3 +1,4 @@
+import sys
 import pygame
 from field import Field
 import os

src/tile.py

@@ -4,10 +4,6 @@ from kb import tractor_kb
 import pytholog as pl
 import random
 from config import TILE_SIZE, FREE_TILES
-import torch
-import torchvision.transforms as transforms
-from PIL import Image
-
 
 class Tile(pygame.sprite.Sprite):
 
@@ -30,40 +26,15 @@ class Tile(pygame.sprite.Sprite):
             self.set_type(random_vegetable)
             self.water_level = random.randint(1, 5) * 10
             self.stage = 'planted' # wczesniej to była self.faza = 'posadzono' ale stwierdzilem ze lepiej po angielsku???
-
-            classes = [
-                    "bób", "brokuł", "brukselka", "burak", "cebula",
-                    "cukinia", "dynia", "fasola", "groch", "jarmuż",
-                    "kalafior", "kalarepa", "kapusta", "marchew", 
-                    "ogórek", "papryka", "pietruszka", "pomidor",
-                    "por", "rzepa", "rzodkiewka", "sałata", "seler",
-                    "szpinak", "ziemniak"]
-
-            model = torch.load("veggies_recognition/best_model.pth")
-
-            mean = [0.5322, 0.5120, 0.3696]
-            std = [0.2487, 0.2436, 0.2531]
-
-            image_transforms = transforms.Compose([
-                transforms.Resize((224, 224)),
-                transforms.ToTensor(),
-                transforms.Normalize(torch.Tensor(mean),torch.Tensor(std))
-            ])
-
-            self.prediction = self.predict(model, image_transforms, self.image_path, classes)
-            
-
         else:
             if random.randint(1, 10) % 3 == 0:
                 self.set_type('water')
                 self.water_level = 100
                 self.stage = 'no_plant'
-                self.prediction = 'water'
             else:
                 self.set_type('grass')
                 self.water_level = random.randint(1, 5) * 10
                 self.stage = 'no_plant'
-                self.prediction = 'grass'
         
 
         self.rect = self.image.get_rect()
@@ -72,17 +43,6 @@ class Tile(pygame.sprite.Sprite):
 
     def draw(self, surface):
         self.tiles.draw(surface)
-    
-    def get_random_image_from_folder(self):
-        folder_path = f"veggies_recognition/veggies/testing/{self.type}"
-
-        files = [f for f in os.listdir(folder_path) if os.path.isfile(os.path.join(folder_path, f))]
-        random_file = random.choice(files)
-
-        #image_path = os.path.join(folder_path, random_file)
-        image_path = folder_path + "/" + random_file
-        #print(image_path)
-        return image_path
 
     def set_type(self, type):
         self.type = type
@@ -91,26 +51,9 @@ class Tile(pygame.sprite.Sprite):
         elif self.type == 'water':
             image_path = "images/water.png"
         else:
-            #image_path = f"images/vegetables/{self.type}.png"
+            image_path = f"images/vegetables/{self.type}.png"
-            image_path = self.get_random_image_from_folder()
             if not os.path.exists(image_path):
                 image_path = "images/question.jpg"
         
-        self.image_path = image_path
         self.image = pygame.image.load(image_path).convert()
         self.image = pygame.transform.scale(self.image, (TILE_SIZE, TILE_SIZE))
-
-    def predict(self, model, image_transforms, image_path, classes):
-        model = model.eval()
-        image = Image.open(image_path)
-        image = image.convert("RGB")
-        image = image_transforms(image).float()
-        image = image.unsqueeze(0)
-
-        output = model(image)
-        _, predicted = torch.max(output.data, 1)
-
-        #print("Rozpoznano: ", classes[predicted.item()])
-        return classes[predicted.item()]
-
-

src/tractor.py

@@ -67,9 +67,7 @@ class Tractor(pygame.sprite.Sprite):
                 neighbors.append('grass')
 
         input_data = {
-            #tutaj będzie dostawał informację ze zdjęcia
+            'tile_type': self.get_current_tile().type,
-            'tile_type': self.get_current_tile().prediction,
-            #'tile_type': self.get_current_tile().type,
             'water_level': self.get_current_tile().water_level,
             "plant_stage": self.get_current_tile().stage,
             "neighbor_N": neighbors[0],
@@ -182,7 +180,6 @@ class Tractor(pygame.sprite.Sprite):
         if (self.get_current_tile().type != 'grass' or self.get_current_tile().type == 'water'): action = 'move'
         self.prev_action = action
 
-
         match (action):
             case ('move'):
                 pass
@@ -243,12 +240,9 @@ class Tractor(pygame.sprite.Sprite):
                 self.get_current_tile().set_type('ziemniak')
         self.move_2()
         #self.action_index += 1
-        print("Rozpoznano: ", self.get_current_tile().prediction)
+        print(action)
-        print("Co jest faktycznie: ", self.get_current_tile().type)
-        print("\n")
-        
         return
-    
+
     def log_info(self):
         # print on what tile type the tractor is on
         x = self.rect.x // TILE_SIZE

src/veggies_recognition/predict.py

@@ -1,36 +0,0 @@
-# import torch
-# import torchvision.transforms as transforms
-# from PIL import Image
-
-# classes = [
-#         "bób", "brokuł", "brukselka", "burak", "cebula",
-#         "cukinia", "dynia", "fasola", "groch", "jarmuż",
-#         "kalafior", "kalarepa", "kapusta", "marchew", 
-#         "ogórek", "papryka", "pietruszka", "pomidor",
-#         "por", "rzepa", "rzodkiewka", "sałata", "seler",
-#         "szpinak", "ziemniak"]
-
-# model = torch.load("best_model.pth")
-
-# mean = [0.5322, 0.5120, 0.3696]
-# std = [0.2487, 0.2436, 0.2531]
-
-# image_transforms = transforms.Compose([
-#     transforms.Resize((224, 224)),
-#     transforms.ToTensor(),
-#     transforms.Normalize(torch.Tensor(mean),torch.Tensor(std))
-# ])
-
-# def predict(model, image_transforms, image_path, classes):
-#     model = model.eval()
-#     image = Image.open(image_path)
-#     print(image_path)
-#     image = image_transforms(image).float()
-#     image = image.unsqueeze(0)
-
-#     output = model(image)
-#     _, predicted = torch.max(output.data, 1)
-
-#     print(classes[predicted.item()])
-
-# predict(model, image_transforms, "veggies/marchew_118.jpg", classes)