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
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
| import numpy as np
class Conv3x3:
def __init__(self, num_filters, learn_rate=0.01):
self.num_filters = num_filters
self.filters = np.random.randn(num_filters, 3, 3) / 9
self.last_input = None
self.learn_rate = learn_rate
def iterate_regions(self, image):
h, w = image.shape
for i in range(h - 2):
for j in range(w - 2):
im_region = image[i:(i + 3), j:(j + 3)]
yield im_region, i, j
def forward(self, input):
self.last_input = input
h, w = input.shape
output = np.zeros((h - 2, w - 2, self.num_filters))
for im_region, i, j in self.iterate_regions(input):
output[i, j] = np.sum(im_region * self.filters, axis=(1, 2))
return output
def backprop(self, d_L_d_out):
d_L_d_filters = np.zeros(self.filters.shape)
for im_region, i, j in self.iterate_regions(self.last_input):
for f in range(self.num_filters):
d_L_d_filters[f] += d_L_d_out[i, j, f] * im_region
self.filters -= self.learn_rate * d_L_d_filters
return None
class MaxPool2:
def __init__(self):
self.last_input = None
def iterate_regions(self, image):
h, w, _ = image.shape
new_h = h // 2
new_w = w // 2
for i in range(new_h):
for j in range(new_w):
im_region = image[i * 2:(i + 1) * 2, j * 2:(j + 1) * 2]
yield im_region, i, j
def forward(self, input):
self.last_input = input
h, w, num_filters = input.shape
output = np.zeros((h // 2, w // 2, num_filters))
for im_region, i, j in self.iterate_regions(input):
output[i, j] = np.amax(im_region, axis=(0, 1))
return output
def backprop(self, d_L_d_out):
d_L_d_input = np.zeros(self.last_input.shape)
for im_region, i, j in self.iterate_regions(self.last_input):
h, w, f = im_region.shape
amax = np.amax(im_region, axis=(0, 1))
for i2 in range(h):
for j2 in range(w):
for f2 in range(f):
if im_region[i2, j2, f2] == amax[f2]:
d_L_d_input[i * 2 + i2, j * 2 + j2, f2] = d_L_d_out[i, j, f2]
return d_L_d_input
class Softmax:
def __init__(self, input_len, nodes, learn_rate=0.01):
self.weights = np.random.randn(input_len, nodes) / nodes
self.biases = np.zeros(nodes)
self.learn_rate = learn_rate
self.last_input_shape = None
self.last_input = None
self.last_totals = None
def forward(self, input):
self.last_input_shape = input.shape
input = input.flatten()
self.last_input = input
totals = np.dot(input, self.weights) + self.biases
self.last_totals = totals
exp = np.exp(totals)
return exp / np.sum(exp, axis=0)
def backprop(self, d_L_d_out):
d_L_d_w = np.zeros(self.weights.shape)
d_L_d_b = np.zeros(self.biases.shape)
d_L_d_input = np.zeros(self.last_input.shape)
for i, gradient in enumerate(d_L_d_out):
if gradient == 0:
continue
t_exp = np.exp(self.last_totals)
S = np.sum(t_exp)
d_out_d_t = -t_exp[i] * t_exp / (S ** 2)
d_out_d_t[i] = t_exp[i] * (S - t_exp[i]) / (S ** 2)
d_t_d_w = self.last_input
d_t_d_b = 1
d_t_d_input = self.weights
d_L_d_t = gradient * d_out_d_t
d_L_d_w += d_t_d_w[np.newaxis].T @ d_L_d_t[np.newaxis]
d_L_d_b += d_L_d_t * d_t_d_b
d_L_d_input += d_t_d_input @ d_L_d_t
self.weights -= self.learn_rate * d_L_d_w
self.biases -= self.learn_rate * d_L_d_b
return d_L_d_input.reshape(self.last_input_shape)
import pickle
class ModelSaver:
def __init__(self, model_name='MNIST_CNN'):
self.model_name = model_name
def save(self, conv, pool, softmax):
data = {
'conv_filters': conv.filters,
'softmax_weights': softmax.weights,
'softmax_biases': softmax.biases
}
filename = f'{self.model_name}.pkl'
with open(filename, 'wb') as f:
pickle.dump(data, f)
print(f"保存参数到{filename}")
def load(self, conv, pool, softmax):
filename = f'{self.model_name}.pkl'
try:
with open(filename, 'rb') as f:
data = pickle.load(f)
conv.filters = data['conv_filters']
softmax.weights = data['softmax_weights']
softmax.biases = data['softmax_biases']
print("模型参数加载成功")
return True
except FileNotFoundError:
print("无可用模型参数")
return False
import tkinter as tk
from PIL import Image,ImageDraw,ImageOps
class DrawingBoard:
def __init__(self, root):
self.root = root
self.root.title("画板")
self.canvas = tk.Canvas(root, width=280, height=280, bg='white')
self.canvas.pack()
self.canvas.bind("<B1-Motion>", self.paint)
self.image = Image.new("RGB", (280, 280), "white")
self.draw = ImageDraw.Draw(self.image)
self.brush_color = "black"
self.brush_width = 5
self.output_button = tk.Button(root, text="输出", command=self.output_and_exit)
self.output_button.pack()
def paint(self, event):
x1, y1 = (event.x - self.brush_width), (event.y - self.brush_width)
x2, y2 = (event.x + self.brush_width), (event.y + self.brush_width)
self.canvas.create_oval(x1, y1, x2, y2, fill=self.brush_color, outline=self.brush_color)
self.draw.ellipse([x1, y1, x2, y2], fill=self.brush_color, outline=self.brush_color)
def process_image(self):
processed_image = self.image.resize((28, 28), Image.Resampling.LANCZOS)
processed_image = ImageOps.grayscale(processed_image)
image_array = np.array(processed_image)
image_array = image_array.astype(np.uint8)
return image_array
def output_and_exit(self):
self.image_array = self.process_image()
processed_image = Image.fromarray(self.image_array)
processed_image.save("temp.png")
print("图片已保存为 temp.png")
self.root.destroy()
|
