-
Notifications
You must be signed in to change notification settings - Fork 0
/
predict.py
42 lines (34 loc) · 1.76 KB
/
predict.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
import numpy as np
import load
labels_map = {
0: "A", 1: "B", 2: "C", 3: "D", 4: "E", 5: "F",
6: "G", 7: "H", 8: "I", 9: "K", 10: "L", 11: "M",
12: "N", 13: "O", 14: "P", 15: "Q", 16: "R", 17: "S",
18: "T", 19: "U", 20: "V", 21: "W", 22: "X", 23: "Y",
}
def cnn_argmax(prediction):
pred = prediction.cpu()
pred = pred.detach().numpy() # converts to <class 'numpy.ndarray'>
maxIdx = np.argmax(pred, axis=1) # returns index of the max value
score = np.amax(np.exp(pred)/np.sum(np.exp(pred))) # returns exponentiated normailized max value
#score = np.amax(prediction) # returns max value
label = labels_map[maxIdx[0]] # returns the label
return label, score
def bcnn_argmax(prediction):
pred = prediction.cpu()
pred = pred.detach().numpy() # converts to <class 'numpy.ndarray'>
maxIdx = np.argmax(pred, axis=1) # returns index of the max value
score = np.amax(np.exp(pred)) # returns exponentiated max value
#score = np.amax(prediction) # returns max value
label = labels_map[maxIdx[0]] # returns the label
return label, score
def CNN(image_data):
cnn = load.CNN() # load pre-trained model
prediction = cnn(image_data) # make prediction using input image
label, score = cnn_argmax(prediction=prediction) # get the most likely class
return label, score
def BCNN(image_data, nsamples):
bcnn = load.BCNN() # load pre-trained model
prediction = bcnn.predict(image_data, num_predictions=nsamples) # make prediction using input image
label, score = bcnn_argmax(prediction) # get the most likely class
return label, score