-
Notifications
You must be signed in to change notification settings - Fork 0
/
temp.py
234 lines (204 loc) · 7.65 KB
/
temp.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
# import torch
# from torch.autograd import Variable
# import numpy as np
# import matplotlib.pyplot as plt
# def coords_fmap2orig(feature,stride):
# '''
# # transfor one fmap coords to orig coords
# # Args
# # featurn [batch_size,h,w,c]
# # stride int
# # Returns
# # coords [n,2]
# '''
# h,w=feature.shape[1:3]
# shifts_x = torch.arange(0, w * stride, stride, dtype=torch.float32)
# shifts_y = torch.arange(0, h * stride, stride, dtype=torch.float32)
# shift_y, shift_x = torch.meshgrid(shifts_y, shifts_x)
# shift_x = torch.reshape(shift_x, [-1])-0.5
# shift_y = torch.reshape(shift_y, [-1])-0.5
# coords = torch.stack([shift_x, shift_y], -1) + stride // 2
# return coords
# feature=torch.Tensor(1,3,3,3)
# print(feature.shape)
# coords=coords_fmap2orig(feature,2)
# print(coords)
# # # temp=torch.rand(2,5)
# # # a=torch.tensor([[1,3,4,3,2],[1,3,4,3,2]])
# # # b=torch.zeros((3,5),dtype=torch.long).scatter_(0,a,temp)
# # # # l=[]
# # # # l.append(a)
# # # # l.append(a)
# # # # print(l)
# # # score_seq = [(-score).argsort() for index, score in enumerate(l)]
# # # # for index, score in enumerate(a):
# # # # print(index,score)
# # # # pred_boxes = [sample_boxes[mask] for sample_boxes, mask in zip(pred_boxes, score_seq)]
# # # # pred_labels = [sample_boxes[mask] for sample_boxes, mask in zip(pred_labels, score_seq)]
# # # pred_scores = [sample_boxes[mask] for sample_boxes, mask in zip(l, score_seq)]
# # # print(a)
# # # print(b)
# # # x = torch.Tensor([16,55])
# # # x=Variable(x,requires_grad=True)
# # # y=x*x
# # # y.mean()
# # # print(y)
# # # # y.backward()
# # # print(y.mean())
# # # target_pos=5
# # # target_pos=(torch.arange(1,10+1)[None,:]==target_pos).float()
# # # print(target_pos)
# # #tensor([[0.1940, 0.3340, 0.8184, 0.4269, 0.5945],
# # # [0.2078, 0.5978, 0.0074, 0.0943, 0.0266]])
# # # b=torch.tensor([[0, 1, 2, 0, 0], [2, 0, 0, 1, 2]])
# # # b=torch.zeros_like(x,dtype=torch.long).scatter_(-1, b.unsqueeze(dim=-1), 1)
# # # b1=b>0
# # # print(x)
# # # print(b)
# # # print(b1)
# # # print(b.shape)
# # # for sample_boxes, mask in zip(l, score_seq):
# # # print(sample_boxes,mask)
# # def sort_by_score(pred_boxes, pred_labels, pred_scores):
# # score_seq = [(-score).argsort() for index, score in enumerate(pred_scores)]
# # pred_boxes = [sample_boxes[mask] for sample_boxes, mask in zip(pred_boxes, score_seq)]
# # pred_labels = [sample_boxes[mask] for sample_boxes, mask in zip(pred_labels, score_seq)]
# # pred_scores = [sample_boxes[mask] for sample_boxes, mask in zip(pred_scores, score_seq)]
# # return pred_boxes, pred_labels, pred_scores
# # # a=np.zeros((0,))
# # # print(a)
# # # a=np.append(a,1)
# # # a=np.append(a,1)
# # def _compute_ap(recall, precision):
# # """ Compute the average precision, given the recall and precision curves.
# # Code originally from https://github.com/rbgirshick/py-faster-rcnn.
# # # Arguments
# # recall: The recall curve (list).
# # precision: The precision curve (list).
# # # Returns
# # The average precision as computed in py-faster-rcnn.
# # """
# # # correct AP calculation
# # # first append sentinel values at the end
# # mrec = np.concatenate(([0.], recall, [1.]))
# # print(mrec)
# # print('\n')
# # mpre = np.concatenate(([0.], precision, [0.]))
# # print(mpre)
# # print('\n')
# # # compute the precision envelope
# # for i in range(mpre.size - 1, 0, -1):
# # mpre[i - 1] = np.maximum(mpre[i - 1], mpre[i])
# # # print(mpre[i-1])
# # print(mpre)
# # print('\n')
# # # to calculate area under PR curve, look for points
# # # where X axis (recall) changes value
# # i = np.where(mrec[1:] != mrec[:-1])[0]
# # print(mrec[1:])
# # print('\n')
# # print( mrec[:-1])
# # print('\n')
# # print(i)
# # plt.figure()
# # plt.plot(mrec,mpre)
# # plt.show()
# # # and sum (\Delta recall) * prec
# # ap = np.sum((mrec[i + 1] - mrec[i]) * mpre[i + 1])
# # return ap
# # tp=np.array([1,1,2,3,3])
# # fp=np.array([0,1,1,2,3])
# # tp=tp/4
# # # print("recall:{}".format(tp))
# # fp=tp/(tp+fp)
# # # print(fp)
# # # print("precision:{}".format(fp.size))
# # _compute_ap(tp,fp)
# # # for i in range(10, 0, -1):
# # # print(i)
# # # pred_boxes=[]
# # # pred_labels=[]
# # # pred_scores=[]
# # # a=torch.tensor([[213,324,344,633],[123,324,344,633],[567,324,344,633]])
# # # label=torch.tensor([2,2,3])
# # # score=torch.tensor([0.8,0.5,0.9])
# # # for _ in range(2):
# # # pred_boxes.append(a)
# # # pred_labels.append(label)
# # # pred_scores.append(score)
# # # # print(pred_scores)
# # # # print(pred_boxes)
# # # out1,out2,out3=sort_by_score(pred_boxes,pred_labels,pred_scores)
# # # print(out2)
# # # p=[simple==2 for simple in out2]
# # # print(p)
# # # gt_single_cls = [sample_boxes[mask] for sample_boxes, mask in zip(out1, p)]
# # # print(gt_single_cls)
# # a=np.array([0.,0.125 ,0.125, 0.25 , 0.375, 0.375 ,0.5 , 0.5 , 0.5 , 0.625 ,1. ])
# # b=np.array([1. , 1. , 0.2 , 0.2 , 0.15789474, 0.11111111,
# # 0.11111111 ,0.09090909, 0.08196721, 0.08196721 ,0. ])
# import pandas as pd
# import numpy as np
# import matplotlib.pyplot as plt
# from matplotlib.patches import PathPatch
# df = pd.DataFrame(np.random.rand(140, 4), columns=['A', 'B', 'C', 'D'])
# df['models'] = pd.Series(np.repeat(['model1','model2', 'model3', 'model4', 'model5', 'model6', 'model7'], 20))
# bp_dict = df.boxplot(
# by="models",layout=(4,1),figsize=(6,8),
# return_type='both',
# patch_artist = True,
# )
# colors = ['b', 'y', 'm', 'c', 'g', 'b', 'r', 'k', ]
# for row_key, (ax,row) in bp_dict.iteritems():
# ax.set_xlabel('')
# for i,box in enumerate(row['boxes']):
# box.set_facecolor(colors[i])
# plt.show()
# def quicksort(arr):
# if len(arr) <= 1:
# return arr
# pivot = arr[len(arr) // 2]
# left = [x for x in arr if x < pivot]
# middle = [x for x in arr if x == pivot]
# right = [x for x in arr if x > pivot]
# return quicksort(left) + middle + quicksort(right)
# print(quicksort([3, 6, 8, 19, 1, 5])) # [1,3, 5, 6, 8, 19
import torch
import torch.nn.functional as f
a=torch.rand(3,4)
tmp=[[0],[0],[0]]
t=torch.tensor(tmp)
b=(torch.arange(1,5,device=a.device)[None,:]==t).long()
loss=torch.nn.CrossEntropyLoss()
loss1=loss(a,t.squeeze(-1))
print(loss1)
print(b)
# import matplotlib.pyplot as plt
# import numpy as np
# import torch
# import torch
# import torch.nn as nn
# loss = nn.CrossEntropyLoss()
# # input, NxC=2x3
# # input = torch.randn((1,2,2,2))
# # print(torch.nn.functional.softmax(input))
# # # target, N
# # target = torch.empty((1,2,2), dtype=torch.long).random_(2)
# # print(target)
# # output = loss(input, target)
# # print(output)
# tensor1=torch.randint(3,13,size=(2,2),dtype=torch.float32,requires_grad=True)
# print(tensor1.unsqueeze(dim=1).shape)
# tensor2=torch.cat((tensor1.unsqueeze(dim=1),tensor1.unsqueeze(dim=1)),dim=1)
# # tensor2.requires_grad=True
# tensor3=tensor1+tensor2
# print(torch.arange(1,10))
#output.backward()
# fig2=plt.figure()
# ax1 = fig1.add_subplot(1,1,1) # 画2行1列个图形的第1个
# ax2 = fig2.add_subplot(1,1,1) # 画2行1列个图形的第2个
# ax1.set_xlabel("efef")
# ax1.plot(np.random.randint(1,5,5), np.arange(5))
# ax2.plot(np.arange(10)*3, np.arange(10))
# fig2.savefig("dfd.png",dpi=3000,bbox_inches='tight')
# plt.show()