Merge pull request 'drzewo impl klasy do podlewania' (#6) from tree_classes into master

Reviewed-on: #6

.idea/misc.xml

@@ -1,4 +1,4 @@
 <?xml version="1.0" encoding="UTF-8"?>
 <project version="4">
-  <component name="ProjectRootManager" version="2" languageLevel="JDK_19" project-jdk-name="Python 3.9" project-jdk-type="Python SDK" />
+  <component name="ProjectRootManager" version="2" project-jdk-name="Python 3.10 (pythonProject)" project-jdk-type="Python SDK" />
 </project>

.vs/VSWorkspaceState.json

@@ -1,8 +1,7 @@
 {
   "ExpandedNodes": [
-    "",
-    "\\decisionTree"
+    ""
   ],
-  "SelectedNode": "\\decisionTree\\treemaker.py",
+  "SelectedNode": "\\C:\\Users\\zmysz\\Desktop\\nowy-inteligentny-traktor",
   "PreviewInSolutionExplorer": false
 }

NN/Generator.py

@@ -0,0 +1,82 @@
+from PIL import Image
+import random
+
+plants = [[], [], []]
+plants[0].append(Image.open("w1.png"))
+plants[0].append(Image.open("w2.png"))
+plants[0].append(Image.open("w3.png"))
+plants[1].append(Image.open("c1.png"))
+plants[1].append(Image.open("c2.png"))
+plants[1].append(Image.open("c3.png"))
+plants[2].append(Image.open("ca1.png"))
+plants[2].append(Image.open("ca2.png"))
+plants[2].append(Image.open("ca3.png"))
+b = [Image.open("b1.png").convert('RGBA'), Image.open("b2.png").convert('RGBA'), Image.open("b3.png").convert('RGBA')]
+
+
+def generate(water, fertilizer, plantf):
+    if water == 1:
+        new_im = Image.new('RGB', (100, 100),
+                           (160 + random.randint(-10, 10), 80 + random.randint(-10, 10), 40 + random.randint(-10, 10)))
+        tmp = plants[plantf][random.randint(0, 2)].resize(
+            (25 + random.randint(-10, 25), 25 + random.randint(-10, 25))).rotate(random.randint(0, 359))
+        new_im.paste(tmp, (random.randint(0, 50), random.randint(0, 50)), tmp)
+        if fertilizer:
+            tmp = b[random.randint(0, 2)].resize(
+                (20 + random.randint(0, 25), 20 + random.randint(0, 25))).rotate(random.randint(0, 359))
+            new_im.paste(tmp, (random.randint(25, 75), random.randint(25, 75)), tmp)
+    else:
+        if fertilizer:
+            new_im = Image.new('RGB', (100, 100),
+                               (
+                                   50 + random.randint(-10, 10), 25 + random.randint(-10, 10),
+                                   0 + random.randint(-10, 10)))
+            tmp = plants[plantf][random.randint(0, 2)].resize(
+                (25 + random.randint(-10, 25), 25 + random.randint(-10, 25))).rotate(random.randint(0, 359))
+            new_im.paste(tmp, (random.randint(0, 50), random.randint(0, 50)), tmp)
+            tmp = b[random.randint(0, 2)].resize(
+                (20 + random.randint(0, 25), 20 + random.randint(0, 25))).rotate(random.randint(0, 359))
+            new_im.paste(tmp, (random.randint(25, 75), random.randint(25, 75)), tmp)
+
+        else:
+            if random.randint(0, 1) == 1:
+                new_im = Image.new('RGB', (100, 100),
+                                   (50 + random.randint(-10, 10), 25 + random.randint(-10, 10),
+                                    0 + random.randint(-10, 10)))
+            else:
+                new_im = Image.new('RGB', (100, 100),
+                                   (160 + random.randint(-10, 10), 80 + random.randint(-10, 10),
+                                    40 + random.randint(-10, 10)))
+            if random.randint(0, 1) == 1:  # big
+                tmp = plants[plantf][random.randint(0, 2)].resize(
+                    (75 + random.randint(-10, 25), 75 + random.randint(-10, 25))).rotate(random.randint(0, 359))
+                new_im.paste(tmp, (random.randint(0, 15), random.randint(0, 15)), tmp)
+            else:
+                tmp = plants[plantf][random.randint(0, 2)].resize(
+                    (random.randint(10, 80), random.randint(10, 80))).rotate(random.randint(0, 359))
+                datas = tmp.getdata()
+
+                new_image_data = []
+                for item in datas:
+                    # change all white (also shades of whites) pixels to yellow
+                    if item[0] in list(range(190, 256)):
+                        new_image_data.append(
+                            (random.randint(0, 10), 255 + random.randint(-150, 0), random.randint(0, 10)))
+                    else:
+                        new_image_data.append(item)
+
+                # update image data
+                tmp.putdata(new_image_data)
+                new_im.paste(tmp, (random.randint(0, 30), random.randint(0, 30)), tmp)
+
+    return new_im
+
+
+for x in range(0, 1000):
+    generate(0, 0, random.randint(0, 2)).save('datasets/00/' + str(x) + '.png')
+for x in range(0, 1000):
+    generate(1, 0, random.randint(0, 2)).save('datasets/10/' + str(x) + '.png')
+for x in range(0, 1000):
+    generate(0, 1, random.randint(0, 2)).save('datasets/01/' + str(x) + '.png')
+for x in range(0, 1000):
+    generate(1, 1, random.randint(0, 2)).save('datasets/11/' + str(x) + '.png')

NN/trainer.py

@@ -0,0 +1,48 @@
+import pathlib
+import random
+
+import torch
+from torch import nn
+from torch.utils.data import DataLoader
+from torchvision import datasets, transforms
+from torchvision.transforms import Lambda
+
+
+device = torch.device('cpu')
+
+
+def train(model, dataset, n_iter=100, batch_size=2560000):
+    optimizer = torch.optim.SGD(model.parameters(), lr=0.01)
+    criterion = nn.NLLLoss()
+    dl = DataLoader(dataset, batch_size=batch_size)
+    model.train()
+    for epoch in range(n_iter):
+        for images, targets in dl:
+            optimizer.zero_grad()
+            out = model(images.to(device))
+            loss = criterion(out, targets.to(device))
+            loss.backward()
+            optimizer.step()
+        if epoch % 10 == 0:
+            print('epoch: %3d loss: %.4f' % (epoch, loss))
+
+
+image_path_list = list(pathlib.Path('./').glob("*/*/*.png"))
+
+random_image_path = random.choice(image_path_list)
+data_transform = transforms.Compose([
+    transforms.Resize(size=(100, 100)),
+    transforms.RandomHorizontalFlip(p=0.5),
+    transforms.ToTensor(),
+    Lambda(lambda x: x.flatten())
+])
+
+train_data = datasets.ImageFolder(root="./datasets",
+                                  transform=data_transform,
+                                  target_transform=None)
+
+model1 = nn.Sequential(nn.Linear(30000, 10000), nn.ReLU(), nn.Linear(10000, 10000), nn.ReLU(), nn.Linear(10000, 0000), nn.Linear(10000, 4), nn.LogSoftmax(dim=-1)).to(device)
+model1.load_state_dict(torch.load("./trained"))
+train(model1, train_data)
+
+torch.save(model1.state_dict(), "./trained")

astar.py

@@ -1,7 +1,6 @@
 from operator import itemgetter
 import cart
 import copy
-from classes import Field
 
 
 class Istate:
@@ -68,23 +67,23 @@ class Node:
         self.y = y
 
 
-def fieldCost(T,node):
+def fieldCost(T, node):
     c = 0
     if T[node.x-1][node.y-1].plantType == 1:
-        c =2
+        c = 2
     elif T[node.x-1][node.y-1].plantType == 2:
-        c =5
+        c = 5
     elif T[node.x-1][node.y-1].plantType == 3:
-        c =13
+        c = 13
     elif T[node.x-1][node.y-1].plantType == 4:
-        c =100000
+        c = 100000
     else:
-        c=0
+        c = 0
 
     if T[node.x-1][node.y-1].isWet == 1:
         c = c + 4
     else:
-        c=c+1
+        c = c+1
 
     return c
 
@@ -92,7 +91,7 @@ def fieldCost(T,node):
 def cost(T, node):
     cost = 0
 
-    while (node.get_parent() != None):
+    while node.get_parent() is not None:
         cost = cost + fieldCost(T, node)
         node = node.get_parent()
 
@@ -103,7 +102,7 @@ def f(goaltest, map, node):
     return cost(map, node) + heuristic(goaltest, node)
 
 
-def goal_test(elem,goaltest):
+def goal_test(elem, goaltest):
     if elem.get_x() == goaltest[0] and elem.get_y() == goaltest[1]:
         return True
     else:
@@ -132,7 +131,7 @@ def graphsearch(explored, f, fringe, goaltest, istate, map, succ):  # przeszukiw
                 explored_tuple.append((x.get_direction(), x.get_x(), x.get_y()))
             x = Node(action, state[0], elem[0], state[1], state[2])  # stworzenie nowego wierzchołka, którego rodzicem jest elem
             p = f(goaltest, map, x)  # liczy priorytet
-            #print('Koszt =', p)
+            # print('Koszt =', p)
             if state not in fringe_tuple and state not in explored_tuple:  # jeżeli stan nie znajduje się na fringe oraz nie znajduje się w liście wierzchołków odwiedzonych
                 fringe.append((x, p))  # dodanie wierzchołka na fringe
                 fringe = sorted(fringe, key=itemgetter(1))  # sortowanie fringe'a według priorytetu
@@ -141,7 +140,7 @@ def graphsearch(explored, f, fringe, goaltest, istate, map, succ):  # przeszukiw
                 for (state_prio, r) in fringe_tuple_prio:
                     if str(state_prio) == str(state):
                         if r > p:
-                            fringe.insert(i, (x,p))  # zamiana state, który należy do fringe z priorytetem r na state z priorytetem p (niższym)
+                            fringe.insert(i, (x, p))  # zamiana state, który należy do fringe z priorytetem r na state z priorytetem p (niższym)
                             fringe.pop(i + 1)
                             fringe = sorted(fringe, key=itemgetter(1))  # sortowanie fringe'a według priorytetu
                             break
@@ -154,7 +153,7 @@ def heuristic(goaltest, node):
 
 def print_moves(elem):
     moves_list = []
-    while (elem.get_parent() != None):
+    while elem.get_parent() is not None:
         moves_list.append(elem.get_action())
         elem = elem.get_parent()
     moves_list.reverse()
@@ -184,7 +183,6 @@ def succ(elem):
     temp_move_east = elem.get_x() + 1
     temp_move_north = elem.get_y() + 1
 
-
     if cart.Cart.is_move_allowed_succ(elem) == "x + 1":
         actions_list.append(("move", (elem.get_direction(), temp_move_east, elem.get_y())))
     elif cart.Cart.is_move_allowed_succ(elem) == "y + 1":
@@ -194,8 +192,4 @@ def succ(elem):
     elif cart.Cart.is_move_allowed_succ(elem) == "x - 1":
         actions_list.append(("move", (elem.get_direction(), temp_move_west, elem.get_y())))
 
-
-
-
-
-    return actions_list
+    return actions_list

bfs.py

@@ -1,4 +1,3 @@
-import sys
 import cart
 import copy
 
@@ -67,23 +66,19 @@ class Node:
         self.y = y
 
 
-def goal_test(goaltest,elem):
+def goal_test(goaltest, elem):
     if elem.get_x() == goaltest[0] and elem.get_y() == goaltest[1]:
         return True
     else:
         return False
 
 
-# def graphsearch(explored, fringe, goaltest, istate, succ):  # przeszukiwanie grafu wszerz
-def graphsearch(explored, fringe, goaltest, istate):  # przeszukiwanie grafu wszerz
+def graphsearch(goaltest, istate):  # przeszukiwanie grafu wszerz
     node = Node(None, istate.get_direction(), None, istate.get_x(), istate.get_y())
 
     fringe = []
-    #elem = []
     explored = []
-    #action = []
     fringe.append(node)  # wierzchołki do odwiedzenia
-    # fringe = [node]
     while True:
         if not fringe:
             return False
@@ -109,7 +104,7 @@ def graphsearch(explored, fringe, goaltest, istate):  # przeszukiwanie grafu wsz
 
 def print_moves(elem):
     moves_list = []
-    while (elem.get_parent() != None):
+    while elem.get_parent() is not None:
         moves_list.append(elem.get_action())
         elem = elem.get_parent()
     moves_list.reverse()
@@ -139,7 +134,6 @@ def succ(elem):
     temp_move_east = elem.get_x() + 1
     temp_move_north = elem.get_y() + 1
 
-
     if cart.Cart.is_move_allowed_succ(elem) == "x + 1":
         actions_list.append(("move", (elem.get_direction(), temp_move_east, elem.get_y())))
     elif cart.Cart.is_move_allowed_succ(elem) == "y + 1":
@@ -149,8 +143,4 @@ def succ(elem):
     elif cart.Cart.is_move_allowed_succ(elem) == "x - 1":
         actions_list.append(("move", (elem.get_direction(), temp_move_west, elem.get_y())))
 
-
-
-
-
-    return actions_list
+    return actions_list

board.py

@@ -1,6 +1,5 @@
 import pygame
 from screen import SCREEN
-global BLACK
 
 # global SCREEN
 global BLACK
@@ -8,9 +7,9 @@ global gridObjects
 global imgTree
 global imgTree
 imgTree = pygame.image.load('img/tree.png')
-
 gridObjects = {}  # Store grid-box objects from Grid Class
 
+
 class Grid(object):
     #  ta klasa rysuje kratę na ekranie
     def __init__(self, x, y, sx, sy):
@@ -27,6 +26,7 @@ class Grid(object):
         BLACK = (0, 0, 0)
         pygame.draw.rect(SCREEN, BLACK, (self.x, self.y, self.sx, self.sy), self.width)
 
+
 class Box(object):
     # global SCREEN
 
@@ -43,6 +43,7 @@ class Box(object):
         # global BLACK
         pygame.draw.rect(SCREEN, self.color, pygame.Rect(self.x, self.y, self.sx, self.sy))
 
+
 class Obstacle(object):
     def __init__(self, mouseObj):
         self.mseX = mouseObj[0]
@@ -54,8 +55,8 @@ class Obstacle(object):
             self.y = g.y
             self.sx = g.sx
             self.sy = g.sy
-            if self.mseX > self.x and self.mseX < self.x + self.sx:
-                if self.mseY > self.y and self.mseY < self.y + self.sy:
+            if self.x < self.mseX < self.x + self.sx:
+                if self.y < self.mseY < self.y + self.sy:
                     self.posX = self.x
                     self.posY = self.y
                     self.gridBox = grid
@@ -66,6 +67,7 @@ class Obstacle(object):
         SCREEN.blit(imgTree, (self.posX, self.posY))
         # pygame.display.update()
 
+
 def getGridBoxes(grid_box):
-       global gridObjects
-       return gridObjects[grid_box]
+    global gridObjects
+    return gridObjects[grid_box]

cart.py

@@ -25,9 +25,7 @@ class Cart:
     def set_y(self, y):
         self.y = y
 
-
-    def is_move_allowed(self,
-                        cart_rect):  # sprawdza czy dany ruch, który chce wykonać wózek jest możliwy, zwraca prawdę lub fałsz
+    def is_move_allowed(self, cart_rect):  # sprawdza czy dany ruch, który chce wykonać wózek jest możliwy, zwraca prawdę lub fałsz
         if self.direction == definitions.CART_DIRECTION_EAST and cart_rect.x + definitions.BLOCK_SIZE < definitions.WIDTH_MAP:
             return True
         elif self.direction == definitions.CART_DIRECTION_SOUTH and cart_rect.y - definitions.BLOCK_SIZE > 0:
@@ -74,4 +72,3 @@ class Cart:
             self.direction = 1
         else:
             self.direction = self.direction + 1
-

decisionTree/generator.py

@@ -1,5 +1,6 @@
 import random
 
+
 # Generowanie unikalnej losowej linii tekstu
 def generate_unique_line(existing_lines):
     while True:
@@ -9,6 +10,7 @@ def generate_unique_line(existing_lines):
         if line not in existing_lines:
             return line
 
+
 # Generowanie 200 unikalnych linii tekstu
 lines = []
 while len(lines) < 200:
@@ -18,4 +20,4 @@ while len(lines) < 200:
 # Zapisywanie linii tekstu do pliku
 with open('decisionTree/database.txt', 'w') as file:
     for line in lines:
-        file.write(line + '\n')
+        file.write(line + '\n')

decisionTree/treemaker.py

@@ -53,7 +53,7 @@ with open("database.txt", 'r') as f:
         view.append(x)
         X1.append(test_list)
 
-f = open("learning_set.txt", "w")              # zapisuje atrybuty s?ownie
+f = open("learning_set.txt", "w")              # zapisuje atrybuty s這wnie
 for i in view:
     f.write(str(i)+"\n")
 f.close()

definitions.py

@@ -1,5 +1,3 @@
-# definicje
-import os
 import pygame
 
 pygame.init()
@@ -21,4 +19,4 @@ HEIGHT_AMOUNT, WIDTH_AMOUNT = 11, 11
 HEIGHT_MAP, WIDTH_MAP = BLOCK_SIZE * HEIGHT_AMOUNT, BLOCK_SIZE * WIDTH_AMOUNT
 HEIGHT, WIDTH = HEIGHT_MAP + BLOCK_SIZE, WIDTH_MAP
 IMAGE_SIZE_NEURAL_NETWORK = 16
-WINDOW = pygame.display.set_mode((WIDTH, HEIGHT))
+WINDOW = pygame.display.set_mode((WIDTH, HEIGHT))

genetic_algorithm.py

@@ -0,0 +1,90 @@
+import random
+import math
+
+
+def create_initial_population(population_size, new_list, player):
+    population = []
+    for _ in range(population_size):
+        chromosome = new_list.copy()
+        chromosome.remove((player.x+1, player.y+1))
+        random.shuffle(chromosome)
+        chromosome.insert(0, (player.x+1, player.y+1))
+        population.append(chromosome)
+    return population
+
+
+def calculate_distance(node1, node2):
+    x1, y1 = node1
+    x2, y2 = node2
+    distance = math.sqrt((x2 - x1) ** 2 + (y2 - y1) ** 2)
+    return distance
+
+
+def calculate_fitness(individual):
+    total_distance = 0
+    num_nodes = len(individual)
+
+    for i in range(num_nodes - 1):
+        node1 = individual[i]
+        node2 = individual[i + 1]
+        distance = calculate_distance(node1, node2)
+
+        total_distance += distance
+
+    if total_distance == 0:
+        fitness = float('inf')
+        return fitness
+
+    fitness = 1 / total_distance
+    return fitness
+
+
+def crossover(parent1, parent2, player):
+    child = [(player.x+1, player.y+1)] + [None] * (len(parent1) - 1)
+    start_index = random.randint(1, len(parent1) - 1)
+    end_index = random.randint(start_index + 1, len(parent1))
+    child[start_index:end_index] = parent1[start_index:end_index]
+    remaining_nodes = [node for node in parent2 if node not in child]
+    child[1:start_index] = remaining_nodes[:start_index - 1]
+    child[end_index:] = remaining_nodes[start_index - 1:]
+    return child
+
+
+def mutate(individual, mutation_rate):
+    for i in range(1, len(individual)):
+        if random.random() < mutation_rate:
+            j = random.randint(1, len(individual) - 1)
+            individual[i], individual[j] = individual[j], individual[i]
+    return individual
+
+
+def genetic_algorithm(new_list, player):
+    max_generations = 200
+    population_size = 200
+    mutation_rate = 0.1
+    population = create_initial_population(population_size, new_list, player)
+    best_individual = None
+    best_fitness = float('-inf')
+
+    for generation in range(max_generations):
+        fitness_values = [calculate_fitness(individual) for individual in population]
+        population = [x for _, x in sorted(zip(fitness_values, population), reverse=True)]
+        fitness_values.sort(reverse=True)
+        best_individuals = population[:10]
+        new_population = best_individuals.copy()
+
+        while len(new_population) < population_size:
+            parent1, parent2 = random.choices(best_individuals, k=2)
+            child = crossover(parent1, parent2, player)
+            child = mutate(child, mutation_rate)
+            new_population.append(child)
+
+        for individual in best_individuals:
+            fitness = calculate_fitness(individual)
+            if fitness > best_fitness:
+                best_fitness = fitness
+                best_individual = individual
+
+        population = new_population[:population_size]
+
+    return best_individual

main.py

@@ -413,7 +413,7 @@ def eventHandler(kbdObj, mouseObj):
 
             # Wczytywanie modelu z pliku
             labels = ['Rain', 'Planted', 'Temperature', 'Sun', 'Snow', 'Moisture', 'Rotten', 'Time']
-            loaded_model = joblib.load('decisionTree/decisionTreeFinal.sav')
+            loaded_model = joblib.load('decisionTree/decisionTree.sav')
             sample = W[goalNode[0]-1][goalNode[1]-1]
 
             # Klasyfikacja przy użyciu wczytanego modelu

screen.py

@@ -1,2 +1,2 @@
 import pygame
-SCREEN = pygame.display.set_mode([600,665])
+SCREEN = pygame.display.set_mode([600, 690])