-
Notifications
You must be signed in to change notification settings - Fork 0
/
main_func.py
294 lines (257 loc) · 12.6 KB
/
main_func.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
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
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
#coding:utf-8
import os
import numpy as np
import cv2
from numpy import linalg as LA
import keras
from keras.applications.vgg16 import VGG16
from keras.preprocessing import image
from keras.applications.vgg16 import preprocess_input
from detection_and_localization.detector import init_detector,det
import pdb
wd = '/mnt/disk1/gyl/new_retail_0226/goods_recognition/data/'
if __name__ == "__main__":
image_name = 'P80121-174446.jpg'
im_path = os.path.join(wd,image_name)
image = cv2.imread(im_path)
cfg_file_name = 'yolo-voc-608'
weights_file_name = 'yolo-voc-608_40000'
# --init detector
net, meta = init_detector(model_cfg_name=cfg_file_name, model_weights_name=weights_file_name)
# --detect and localization
res = det(im_path,net,meta,conf_thres=0.2) #[cls,conf,x,y,w,h]
if len(res)==0:
print('No goods detected!')
# pdb.set_trace()
# --recognition
input_shape = [224, 224, 3]
model = VGG16(weights = 'imagenet', input_shape = (input_shape[0], input_shape[1], input_shape[2]), pooling = 'max', include_top = False)
def extract_feat(img, model):
#pdb.set_trace()
input_shape = (224, 224, 3)
#img = image.load_img(img_path, target_size=(input_shape[0], input_shape[1]))
#pdb.set_trace()
img=cv2.resize(img,(input_shape[0], input_shape[1]),interpolation=cv2.INTER_CUBIC)
img = keras.preprocessing.image.img_to_array(img)
#img = np.array(img)
img = np.expand_dims(img, axis=0)
img = preprocess_input(img)
feat = model.predict(img)
norm_feat = feat[0]/LA.norm(feat[0])#也许，这就是归一化把
return norm_feat
#载入各产品类的名称,即groundtruth图片名
name_path='/mnt/disk1/gyl/new_retail_0226/goods_name/'
#number 1
baihujiao_name = open(name_path+'baihujiao.xml').read()
baihujiao_name=baihujiao_name.split()
baihujiao_name_num=len(baihujiao_name)
#number 2
baishiguan_name = open(name_path+'baishiguan.xml').read()
baishiguan_name=baishiguan_name.split()
baishiguan_name_num=len(baishiguan_name)
#number 3
baishiping_name = open(name_path+'baishiping.xml').read()
baishiping_name=baishiping_name.split()
baishiping_name_num=len(baishiping_name)
#number 4
fenda_name = open(name_path+'fenda.xml').read()
fenda_name=fenda_name.split()
fenda_name_num=len(fenda_name)
#number 5
guolicheng_name = open(name_path+'guolicheng.xml').read()
guolicheng_name=guolicheng_name.split()
guolicheng_name_num=len(guolicheng_name)
#number 6
hongshaoniurou_name = open(name_path+'hongshaoniurou.xml').read()
hongshaoniurou_name=hongshaoniurou_name.split()
hongshaoniurou_name_num=len(hongshaoniurou_name)
#number 7
laotansuancaidai_name = open(name_path+'laotansuancaidai.xml').read()
laotansuancaidai_name=laotansuancaidai_name.split()
laotansuancaidai_name_num=len(laotansuancaidai_name)
#number 8
laotansuancaihe_name = open(name_path+'laotansuancaihe.xml').read()
laotansuancaihe_name=laotansuancaihe_name.split()
laotansuancaihe_name_num=len(laotansuancaihe_name)
#number 9
nongfushanquan_name = open(name_path+'nongfushanquan.xml').read()
nongfushanquan_name=nongfushanquan_name.split()
nongfushanquan_name_num=len(nongfushanquan_name)
#number 10
pijiujin_name = open(name_path+'pijiujin.xml').read()
pijiujin_name=pijiujin_name.split()
pijiujin_name_num=len(pijiujin_name)
#number 11
pijiuyin_name = open(name_path+'pijiuyin.xml').read()
pijiuyin_name=pijiuyin_name.split()
pijiuyin_name_num=len(pijiuyin_name)
#number 12
shupiandai_name = open(name_path+'shupiandai.xml').read()
shupiandai_name=shupiandai_name.split()
shupiandai_name_num=len(shupiandai_name)
#number 13
shupiantong_name = open(name_path+'shupiantong.xml').read()
shupiantong_name=shupiantong_name.split()
shupiantong_name_num=len(shupiantong_name)
#number 14
tudouya_name = open(name_path+'tudouya.xml').read()
tudouya_name=tudouya_name.split()
tudouya_name_num=len(tudouya_name)
#number 15
xianggudunji_name = open(name_path+'xianggudunji.xml').read()
xianggudunji_name=xianggudunji_name.split()
xianggudunji_name_num=len(xianggudunji_name)
#number 16
xuebi_name = open(name_path+'xuebi.xml').read()
xuebi_name=xuebi_name.split()
xuebi_name_num=len(xuebi_name)
#读取数据集的所有图片名
all_db_name = open('/mnt/disk1/gyl/new_retail_0226/goods16_name.xml').read()
all_db_name=all_db_name.split()
# read ku image's feature
a = np.loadtxt('/mnt/disk1/gyl/new_retail_0226/goods16_data_2018_02_24_1603.xml')
feats = a.reshape(215,512)
Topk = 1
roi_count = 0
image_size=image.shape
h=image_size[0]
w=image_size[1]
#for控制商品大图image中检测出的所有商品个数，每一个商品自成一个query_image
for roi_num in range(0,len(res)):
#读取query图片.这里的query_image用的就是商品检测的基于 x,y,w,h 的roi区域
#image_size=image.shape
#h=image_size[0]
#w=image_size[1]
#pdb.set_trace()
image_cx = int(res[roi_num][2]*w)
image_cy = int(res[roi_num][3]*h)
image_w = int(res[roi_num][4]*w)
image_h = int(res[roi_num][5]*h)
#变换为roi的左上角坐标image_tf
image_tfx = image_cx - int(image_w/2)
image_tfy = image_cy - int(image_h/2)
#越界判断
if(image_tfx < 0):
image_tfx = 0
if(image_tfy < 0):
image_tfy = 0
if((image_tfx + image_w)>w):
break
if((image_tfy + image_h)>h):
break
query_image = image[image_tfy:image_tfy+image_h,image_tfx:image_tfx+image_w]
#pdb.set_trace()
queryVec = extract_feat(query_image,model)
scores = np.dot(queryVec, feats.T)
scores = np.array(scores)
rank_ID = np.argsort(-scores) #从大到小逆序排序 rank_ID[0]存的就是得分最高图片的序号
rank_score = scores[rank_ID] #已经把相似度按从大到小的顺序排列了，rank_score[0]就是最高的得分，即最相似的图片
top_similar_image=all_db_name[rank_ID[0]] #找到相似图片在文档中对应的图片名
# resultImg = cv2.imread('D:\\Text_image\\Product_Retrieval\\Goods_db\\'+top_similar_image+'.jpg')
# cv2.imshow("resultImg",resultImg)
# cv2.waitKey(0)
print(' -> ',roi_num, top_similar_image,' most similar picture:','%-5d'%(rank_ID[0]+1),'scores:',rank_score[0])
#得分大于0.78才认为是同类商品,下面进行商品归类
if(rank_score[0]>0.70):
for k in range(0,Topk):
#number 1
for j in range(0,baihujiao_name_num):
if(all_db_name[rank_ID[k]]==baihujiao_name[j]):
font=cv2.FONT_HERSHEY_SIMPLEX
cv2.rectangle(image, (image_tfx, image_tfy), (image_tfx+image_w, image_tfy+image_h), (0,255,0),2)
cv2.putText(image,'baihujiao',(image_tfx,image_tfy),font,0.8,(0,175,255),1)
#number 2
for j in range(0,baishiguan_name_num):
if(all_db_name[rank_ID[k]]==baishiguan_name[j]):
font=cv2.FONT_HERSHEY_SIMPLEX
cv2.rectangle(image, (image_tfx, image_tfy), (image_tfx+image_w, image_tfy+image_h), (0,255,0),2)
cv2.putText(image,'baishiguan',(image_tfx,image_tfy),font,0.8,(0,20,255),1)
#number 3
for j in range(0,baishiping_name_num):
if(all_db_name[rank_ID[k]]==baishiping_name[j]):
font=cv2.FONT_HERSHEY_SIMPLEX
cv2.rectangle(image, (image_tfx, image_tfy), (image_tfx+image_w, image_tfy+image_h), (0,255,0),2)
cv2.putText(image,'baishiping',(image_tfx,image_tfy),font,0.8,(0,20,255),1)
#number 4
for j in range(0,fenda_name_num):
if(all_db_name[rank_ID[k]]==fenda_name[j]):
font=cv2.FONT_HERSHEY_SIMPLEX
cv2.rectangle(image, (image_tfx, image_tfy), (image_tfx+image_w, image_tfy+image_h), (0,255,0),2)
cv2.putText(image,'fenda',(image_tfx,image_tfy),font,0.8,(0,20,255),1)
#number 5
for j in range(0,guolicheng_name_num):
if(all_db_name[rank_ID[k]]==guolicheng_name[j]):
font=cv2.FONT_HERSHEY_SIMPLEX
cv2.rectangle(image, (image_tfx, image_tfy), (image_tfx+image_w, image_tfy+image_h), (0,255,0),2)
cv2.putText(image,'guolicheng',(image_tfx,image_tfy),font,0.8,(0,20,255),1)
#number 6
for j in range(0,hongshaoniurou_name_num):
if(all_db_name[rank_ID[k]]==hongshaoniurou_name[j]):
font=cv2.FONT_HERSHEY_SIMPLEX
cv2.rectangle(image, (image_tfx, image_tfy), (image_tfx+image_w, image_tfy+image_h), (0,255,0),2)
cv2.putText(image,'hongshaoniurou',(image_tfx,image_tfy),font,0.8,(0,20,255),1)
#number 7
for j in range(0,laotansuancaidai_name_num):
if(all_db_name[rank_ID[k]]==laotansuancaidai_name[j]):
font=cv2.FONT_HERSHEY_SIMPLEX
cv2.rectangle(image, (image_tfx, image_tfy), (image_tfx+image_w, image_tfy+image_h), (0,255,0),2)
cv2.putText(image,'laotansuancaidai',(image_tfx,image_tfy),font,1.5,(0,20,255),2)
#number 8
for j in range(0,laotansuancaihe_name_num):
if(all_db_name[rank_ID[k]]==laotansuancaihe_name[j]):
font=cv2.FONT_HERSHEY_SIMPLEX
cv2.rectangle(image, (image_tfx, image_tfy), (image_tfx+image_w, image_tfy+image_h), (0,255,0),2)
cv2.putText(image,'laotansuancaihe',(image_tfx,image_tfy),font,1.5,(0,20,255),2)
#number 9
for j in range(0,nongfushanquan_name_num):
if(all_db_name[rank_ID[k]]==nongfushanquan_name[j]):
font=cv2.FONT_HERSHEY_SIMPLEX
cv2.rectangle(image, (image_tfx, image_tfy), (image_tfx+image_w, image_tfy+image_h), (0,255,0),2)
cv2.putText(image,'nongfushanquan',(image_tfx,image_tfy),font,0.8,(0,140,255),1)
#number 10
for j in range(0,pijiujin_name_num):
if(all_db_name[rank_ID[k]]==pijiujin_name[j]):
font=cv2.FONT_HERSHEY_SIMPLEX
cv2.rectangle(image, (image_tfx, image_tfy), (image_tfx+image_w, image_tfy+image_h), (0,255,0),2)
cv2.putText(image,'pijiujin',(image_tfx,image_tfy),font,1.5,(0,20,255),2)
#print("ok - jinse guanyonglai")
#number 11
for j in range(0,pijiuyin_name_num):
if(all_db_name[rank_ID[k]]==pijiuyin_name[j]):
font=cv2.FONT_HERSHEY_SIMPLEX
cv2.rectangle(image, (image_tfx, image_tfy), (image_tfx+image_w, image_tfy+image_h), (0,255,0),2)
cv2.putText(image,'pijiuyin',(image_tfx,image_tfy),font,1.5,(0,20,255),2)
#print("ok - yinse")
#number 12
for j in range(0,shupiandai_name_num):
if(all_db_name[rank_ID[k]]==shupiandai_name[j]):
font=cv2.FONT_HERSHEY_SIMPLEX
cv2.rectangle(image, (image_tfx, image_tfy), (image_tfx+image_w, image_tfy+image_h), (0,255,0),2)
cv2.putText(image,'shupiandai',(image_tfx,image_tfy),font,0.8,(0,20,255),1)
#number 13
for j in range(0,shupiantong_name_num):
if(all_db_name[rank_ID[k]]==shupiantong_name[j]):
font=cv2.FONT_HERSHEY_SIMPLEX
cv2.rectangle(image, (image_tfx, image_tfy), (image_tfx+image_w, image_tfy+image_h), (0,255,0),2)
cv2.putText(image,'shupiantong',(image_tfx,image_tfy),font,0.8,(0,20,255),1)
#number 14
for j in range(0,tudouya_name_num):
if(all_db_name[rank_ID[k]]==tudouya_name[j]):
font=cv2.FONT_HERSHEY_SIMPLEX
cv2.rectangle(image, (image_tfx, image_tfy), (image_tfx+image_w, image_tfy+image_h), (0,255,0),2)
cv2.putText(image,'tudouya',(image_tfx,image_tfy),font,0.8,(0,20,255),1)
#number 15
for j in range(0,xianggudunji_name_num):
if(all_db_name[rank_ID[k]]==xianggudunji_name[j]):
font=cv2.FONT_HERSHEY_SIMPLEX
cv2.rectangle(image, (image_tfx, image_tfy), (image_tfx+image_w, image_tfy+image_h), (0,255,0),2)
cv2.putText(image,'xianggudunji',(image_tfx,image_tfy),font,0.8,(0,20,255),1)
#number 16
for j in range(0,xuebi_name_num):
if(all_db_name[rank_ID[k]]==xuebi_name[j]):
font=cv2.FONT_HERSHEY_SIMPLEX
cv2.rectangle(image, (image_tfx, image_tfy), (image_tfx+image_w, image_tfy+image_h), (0,255,0),2)
cv2.putText(image,'xuebi',(image_tfx,image_tfy),font,0.4,(0,20,255),1)
cv2.imwrite('/mnt/disk1/gyl/'+image_name,image)
print("imwrite picture to /mnt/disk1/gyl/result_reco.jpg is OK!")
# --plot bb on image