Hi every one ,I am working on VGG-Net convolution neural network for multiclassification .I trained this network for face recognition .I used 11 person present 11 class .The label of classes are from 0 to 10.
I used softMax as activation function .My question is :
I selected a picture and tried to recognize the person face in it .The result was a float number ,how can I know to any class this face image belongs to(If I want to test this network manually)?
When I trained the same network for binary classification (face detection) and with using sigmoid activation function I wrote this python code :
result = loaded_model.predict(xx_test)
for i in range(len(xx_test)): # for all the images if result[i][0] >= 0.9:
print("result= " + str(result[i][0]))
prediction = 'not human face'
print("the entered image is ", prediction)
else:
print("result= " + str(result[i][0]))
prediction = 'human face' print("the entered image is ", prediction)
*What about multiclassification what should i write and how i can classify the result ?
Thank you
exaples of the result
3.789948e-08
************
7.274903e-07
************
0.0015600622
************
1.820412e-05
************
4.9111115e-08
************
4.481249e-09
************
7.254938e-06
************
8.680798e-11
Hope I understand your problem correctly.
Use decode_predictions to interpret the results from predict
pred = decode_predictions(result)[0]
print(pred) # should see list of tuples of classes from most prob to least
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Writing