-
Notifications
You must be signed in to change notification settings - Fork 0
/
classifierApp.py
191 lines (153 loc) · 8.51 KB
/
classifierApp.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
import tkinter as tk
from tkinter import filedialog
from PIL import Image, ImageTk
import tensorflow as tf
import numpy as np
import cv2
from tensorflow.keras.models import load_model
from tensorflow.keras.preprocessing.image import img_to_array, load_img
class SatelliteImageClassificationApp(tk.Tk):
def __init__(self):
super().__init__()
self.title("Satellite Image Classification")
self.state('zoomed') # Maximize window to full screen
self.model = None
self.model_path = tk.StringVar(value="No model loaded")
self.init_ui()
def init_ui(self):
self.label_title = tk.Label(self, text="Satellite Image Classification", font=("Helvetica", 20))
self.label_title.pack(pady=20)
self.frame_buttons = tk.Frame(self)
self.frame_buttons.pack(pady=10)
self.btn_select_model = tk.Button(self.frame_buttons, text="Select Model", command=self.load_model, font=("Helvetica", 14), bg="black", fg="white")
self.btn_select_model.pack(side=tk.LEFT, padx=5)
self.btn_select_image = tk.Button(self.frame_buttons, text="Select Image", command=self.load_image, font=("Helvetica", 14), bg="black", fg="white", state=tk.DISABLED)
self.btn_select_image.pack(side=tk.LEFT, padx=5)
self.label_model = tk.Label(self, textvariable=self.model_path, font=("Helvetica", 14))
self.label_model.pack(pady=10)
self.frame = tk.Frame(self)
self.frame.pack(fill=tk.BOTH, expand=True)
self.canvas_original = tk.Canvas(self.frame, bg=self.cget("bg"))
self.canvas_original.pack(side=tk.LEFT, fill=tk.BOTH, expand=True)
self.canvas_colored = tk.Canvas(self.frame, bg=self.cget("bg"))
self.canvas_colored.pack(side=tk.LEFT, fill=tk.BOTH, expand=True)
self.legend_frame = tk.Frame(self, bg=self.cget("bg"))
self.legend_frame.pack(fill=tk.X, pady=10)
self.image_path = None
self.original_image = None
self.colored_image = None
def load_model(self):
model_path = filedialog.askopenfilename(filetypes=[("Keras model files", "*.keras;*.h5")])
if model_path:
self.model = load_model(model_path)
self.model_path.set(f"Current model: {model_path}")
self.btn_select_image.config(state=tk.NORMAL)
# If an image is already selected, process it with the new model
if self.image_path:
self.process_image(self.image_path)
def load_image(self):
self.image_path = filedialog.askopenfilename(filetypes=[("Image files", "*.jpg;*.jpeg;*.png")])
if self.image_path:
self.process_image(self.image_path)
def process_image(self, image_path):
grid_size = 64
class_colors = {
0: [144, 238, 144], # Light Green for AnnualCrop
1: [34, 139, 34], # Dark Green for Forest
2: [173, 255, 47], # Yellow-Green for HerbaceousVegetation
3: [169, 169, 169], # Gray for Highway
4: [211, 211, 211], # Light Gray for Industrial
5: [255, 255, 224], # Light Yellow for Pasture
6: [255, 165, 0], # Light Orange for PermanentCrop
7: [255, 69, 0], # Dark Orange for Residential
8: [0, 0, 255], # Blue for River
9: [135, 206, 235] # Light Blue for SeaLake
}
class_names = [
'AnnualCrop', 'Forest', 'HerbaceousVegetation', 'Highway', 'Industrial',
'Pasture', 'PermanentCrop', 'Residential', 'River', 'SeaLake'
]
# Load and preprocess the real image
real_image = load_img(image_path)
real_image = img_to_array(real_image)
real_image = real_image / 255.0 # Normalize pixel values
# Divide the resized image into grids
grids = self.divide_image_into_grids(real_image, grid_size)
# Predict the class of each grid
predictions = self.model.predict(grids)
predicted_classes = np.argmax(predictions, axis=1)
# Colorize the grids according to the predicted classes
colored_image = self.colorize_grids(real_image, predicted_classes, grid_size, class_colors, real_image.shape[0], real_image.shape[1])
# Resize the colored image back to the original size
self.original_image = real_image
self.colored_image = colored_image
self.display_results(class_names, class_colors)
def divide_image_into_grids(self, image, grid_size):
img_height, img_width, _ = image.shape
grids = []
for y in range(0, img_height, grid_size):
for x in range(0, img_width, grid_size):
grid = image[y:y+grid_size, x:x+grid_size]
if grid.shape[0] != grid_size or grid.shape[1] != grid_size:
grid = cv2.resize(grid, (grid_size, grid_size))
grids.append(grid)
return np.array(grids)
def colorize_grids(self, original_image, predictions, grid_size, class_colors, img_height, img_width):
colored_image = np.zeros_like(original_image)
grid_idx = 0
for y in range(0, img_height, grid_size):
for x in range(0, img_width, grid_size):
if grid_idx < len(predictions):
color = np.array(class_colors[predictions[grid_idx]]) / 255.0
actual_grid_height = min(grid_size, img_height - y)
actual_grid_width = min(grid_size, img_width - x)
resized_color = cv2.resize(np.full((grid_size, grid_size, 3), color), (actual_grid_width, actual_grid_height))
colored_image[y:y+actual_grid_height, x:x+actual_grid_width] = resized_color
# Draw grid lines
cv2.rectangle(colored_image, (x, y), (x + actual_grid_width, y + actual_grid_height), (0, 0, 0), 1)
# Put grid numbers
cv2.putText(colored_image, str(grid_idx), (x + 5, y + 20), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (255, 255, 255), 1)
grid_idx += 1
return colored_image
def display_results(self, class_names, class_colors):
self.canvas_original.delete("all")
self.canvas_colored.delete("all")
for widget in self.legend_frame.winfo_children():
widget.destroy()
# Get image dimensions
img_height, img_width, _ = self.original_image.shape
canvas_width = self.canvas_original.winfo_width()
canvas_height = self.canvas_original.winfo_height()
# Calculate the aspect ratio
aspect_ratio = img_width / img_height
# Calculate dimensions for display
display_width = min(canvas_width, int(canvas_height * aspect_ratio))
display_height = int(display_width / aspect_ratio)
# Convert images to PIL format and resize to fit within the canvas while preserving aspect ratio
original_image_pil = Image.fromarray((self.original_image * 255).astype(np.uint8))
colored_image_pil = Image.fromarray((self.colored_image * 255).astype(np.uint8))
original_image_pil = original_image_pil.resize((display_width, display_height), Image.Resampling.LANCZOS)
colored_image_pil = colored_image_pil.resize((display_width, display_height), Image.Resampling.LANCZOS)
# Convert images to ImageTk format
original_image_tk = ImageTk.PhotoImage(original_image_pil)
colored_image_tk = ImageTk.PhotoImage(colored_image_pil)
# Display original image
self.canvas_original.create_image(0, 0, anchor="nw", image=original_image_tk)
self.canvas_original.image = original_image_tk
# Display colored image
self.canvas_colored.create_image(0, 0, anchor="nw", image=colored_image_tk)
self.canvas_colored.image = colored_image_tk
# Display legend centered
total_legend_width = 0
legend_labels = []
max_height = 2 # set a fixed height for all labels in the legend
for i, (class_name, color) in enumerate(class_colors.items()):
legend_label = tk.Label(self.legend_frame, text=class_names[i], bg=f"#{int(color[0]):02x}{int(color[1]):02x}{int(color[2]):02x}", font=("Helvetica", 12), height=max_height)
legend_label.pack(side=tk.LEFT, padx=5, pady=5)
legend_labels.append(legend_label)
total_legend_width += legend_label.winfo_reqwidth() + 10
# Center the legend frame
self.legend_frame.pack_configure(padx=(self.winfo_width() - total_legend_width) // 2)
if __name__ == '__main__':
app = SatelliteImageClassificationApp()
app.mainloop()