-
Notifications
You must be signed in to change notification settings - Fork 0
/
check_images.txt
31 lines (17 loc) · 1.85 KB
/
check_images.txt
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
Questions regarding Uploaded Image Classification:
1. Did the three model architectures classify the breed of dog in Dog_01.jpg to be the same breed? If not, report the differences in the classifications.
Answer: The three model were able to classify the breed of dog in Dog_01.jpg to be the same breed - basset, basset hound. That's strange n?
As i didn't expected that. I thought that one of the model should fail. :)
2. Did each of the three model architectures classify the breed of dog in Dog_01.jpg to be the same breed of dog as that model architecture classified Dog_02.jpg? If not, report the differences in the classifications.
Answer: None of the three model architectures classified the breed of dog in Dog_01.jpg to be the same breed of dog as that model classified Dog_O2.jpg.
vgg classified Dog_01 as "basset, basset hound", and Dog_02 as "walker hound, walker foxhound".
resnet classified Dog_01 as "basset, basset hound", and Dog_02 as "boston bull, boston terrier".
alexnet classified 01 as "basset, basset hound", and Dog_02 as "siamese cat, siamese, cat".
Alex remind me of my friend.
3. Did the three model architectures correctly classify Animal_Name_01.jpg and Object_Name_01.jpg to not be dogs? If not, report the misclassifications.
Answer: The three model correctly classified Animal_Name_01.jpg and Object_Name_01.jpg to not be dogs.
4. Based upon your answers for questions 1. - 3. above, select the model architecture that you feel did the best at classifying the four uploaded images. Describe why you selected that model architecture as the best on uploaded image classification.
Answer:The model I feel did the best at classifying the four uploaded images is #Vgg. This is because:
1. It classifies the coffee mug as cup.
2. It classified dogs and not dogs better.
3. It's speed for recognizing the images is fast as compared to other models.