Hugging Face's logo

dima806
/
dogs_70_breeds_image_detection

Image Classification
Transformers
PyTorch
Safetensors
vit
Inference Endpoints
Model card Files Community
1
dima806 commited on
Commit
5a36212
·
verified ·
1 Parent(s): e738009

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +75 -76
README.md CHANGED
@@ -7,90 +7,89 @@ base_model:
7
  ---
8
  Predicts dogs breed based on an image.
9
 
10
- Achieved about 90% accuracy on unseen (test) data.
11
 
12
  See [my Kaggle notebook](https://www.kaggle.com/code/dima806/70-dog-breed-image-detection-vit) and [my Medium article](https://medium.com/gitconnected/paws-and-pixels-creating-a-dog-breeds-classifier-with-googles-vision-transformer-431137422830) for more details.
13
 
14
- ![image/png](https://cdn-uploads.huggingface.co/production/uploads/6449300e3adf50d864095b90/0hA9Hg_TgBPqLYAEnIXeK.png)
15
 
16
  ```
17
  Classification report:
18
 
19
  precision recall f1-score support
20
 
21
- Afghan 0.9878 0.9310 0.9586 87
22
- African Wild Dog 1.0000 1.0000 1.0000 87
23
- Airedale 0.9775 1.0000 0.9886 87
24
- American Spaniel 0.5370 1.0000 0.6988 87
25
- American Hairless 0.7711 0.7273 0.7485 88
26
- American Spaniel 1.0000 0.0230 0.0449 87
27
- Basenji 0.9444 0.9770 0.9605 87
28
- Basset 0.8776 0.9885 0.9297 87
29
- Beagle 1.0000 0.8736 0.9325 87
30
- Bearded Collie 0.8788 1.0000 0.9355 87
31
- Bermaise 0.9667 1.0000 0.9831 87
32
- Bichon Frise 0.9451 0.9885 0.9663 87
33
- Blenheim 0.9263 1.0000 0.9617 88
34
- Bloodhound 0.9444 0.9659 0.9551 88
35
- Bluetick 1.0000 0.9885 0.9942 87
36
- Border Collie 0.8500 0.9770 0.9091 87
37
- Borzoi 1.0000 0.9773 0.9885 88
38
- Boston Terrier 0.6014 0.9885 0.7478 87
39
- Boxer 0.9651 0.9432 0.9540 88
40
- Bull Mastiff 0.8913 0.9425 0.9162 87
41
- Bull Terrier 0.9551 0.9770 0.9659 87
42
- Bulldog 1.0000 0.1705 0.2913 88
43
- Cairn 0.8804 0.9310 0.9050 87
44
- Chihuahua 0.9571 0.7701 0.8535 87
45
- Chinese Crested 1.0000 0.9195 0.9581 87
46
- Chow 0.9775 1.0000 0.9886 87
47
- Clumber 0.9551 0.9770 0.9659 87
48
- Cockapoo 0.7209 0.7126 0.7168 87
49
- Cocker 0.9639 0.9195 0.9412 87
50
- Collie 0.9868 0.8621 0.9202 87
51
- Corgi 0.9438 0.9655 0.9545 87
52
- Coyote 0.9167 1.0000 0.9565 88
53
- Dalmation 0.8627 1.0000 0.9263 88
54
- Dhole 1.0000 0.9318 0.9647 88
55
- Dingo 0.8300 0.9540 0.8877 87
56
- Doberman 0.9438 0.9655 0.9545 87
57
- Elk Hound 0.9655 0.9655 0.9655 87
58
- French Bulldog 0.8953 0.8851 0.8902 87
59
- German Sheperd 0.8438 0.9310 0.8852 87
60
- Golden Retriever 0.9545 0.9655 0.9600 87
61
- Great Dane 0.9649 0.6322 0.7639 87
62
- Great Perenees 0.9255 1.0000 0.9613 87
63
- Greyhound 0.9726 0.8161 0.8875 87
64
- Groenendael 0.9529 0.9310 0.9419 87
65
- Irish Spaniel 0.8081 0.9091 0.8556 88
66
- Irish Wolfhound 0.9778 1.0000 0.9888 88
67
- Japanese Spaniel 0.8723 0.9425 0.9061 87
68
- Komondor 0.9767 0.9655 0.9711 87
69
- Labradoodle 0.6040 0.7011 0.6489 87
70
- Labrador 0.9494 0.8621 0.9036 87
71
- Lhasa 0.8621 0.5747 0.6897 87
72
- Malinois 0.9200 0.7931 0.8519 87
73
- Maltese 0.7748 0.9885 0.8687 87
74
- Mex Hairless 0.7273 0.9195 0.8122 87
75
- Newfoundland 0.9647 0.9425 0.9535 87
76
- Pekinese 0.9756 0.9195 0.9467 87
77
- Pit Bull 0.8830 0.9540 0.9171 87
78
- Pomeranian 0.9438 0.9655 0.9545 87
79
- Poodle 0.9444 0.7727 0.8500 88
80
- Pug 0.9651 0.9432 0.9540 88
81
- Rhodesian 0.9121 0.9540 0.9326 87
82
- Rottweiler 0.9560 0.9886 0.9721 88
83
- Saint Bernard 1.0000 0.9655 0.9825 87
84
- Schnauzer 0.9459 0.8046 0.8696 87
85
- Scotch Terrier 0.8391 0.8391 0.8391 87
86
- Shar_Pei 0.8269 0.9885 0.9005 87
87
- Shiba Inu 0.9718 0.7931 0.8734 87
88
- Shih-Tzu 0.7619 0.7356 0.7485 87
89
- Siberian Husky 0.9565 1.0000 0.9778 88
90
- Vizsla 0.9545 0.9655 0.9600 87
91
- Yorkie 0.8947 0.9770 0.9341 87
92
 
93
- accuracy 0.8963 6192
94
- macro avg 0.9099 0.8963 0.8879 6192
95
- weighted avg 0.9099 0.8963 0.8879 6192
96
  ```
 
7
  ---
8
  Predicts dogs breed based on an image.
9
 
10
+ Achieved about 92% accuracy on unseen (test) data.
11
 
12
  See [my Kaggle notebook](https://www.kaggle.com/code/dima806/70-dog-breed-image-detection-vit) and [my Medium article](https://medium.com/gitconnected/paws-and-pixels-creating-a-dog-breeds-classifier-with-googles-vision-transformer-431137422830) for more details.
13
 
14
+ ![image/png](https://cdn-uploads.huggingface.co/production/uploads/6449300e3adf50d864095b90/KdUJH1qZauUr5kargckRa.png)
15
 
16
  ```
17
  Classification report:
18
 
19
  precision recall f1-score support
20
 
21
+ Afghan 0.9753 0.9080 0.9405 87
22
+ African Wild Dog 1.0000 1.0000 1.0000 88
23
+ Airedale 1.0000 0.9885 0.9942 87
24
+ American Hairless 0.9841 0.7126 0.8267 87
25
+ American Spaniel 0.9054 0.7701 0.8323 87
26
+ Basenji 0.9659 0.9770 0.9714 87
27
+ Basset 0.9551 0.9770 0.9659 87
28
+ Beagle 0.9659 0.9659 0.9659 88
29
+ Bearded Collie 0.8614 1.0000 0.9255 87
30
+ Bermaise 0.9457 1.0000 0.9721 87
31
+ Bichon Frise 0.9551 0.9770 0.9659 87
32
+ Blenheim 0.9062 1.0000 0.9508 87
33
+ Bloodhound 0.9659 0.9770 0.9714 87
34
+ Bluetick 1.0000 0.9540 0.9765 87
35
+ Border Collie 0.8830 0.9540 0.9171 87
36
+ Borzoi 1.0000 0.9432 0.9708 88
37
+ Boston Terrier 0.5513 0.9773 0.7049 88
38
+ Boxer 1.0000 0.9655 0.9825 87
39
+ Bull Mastiff 0.9655 0.9655 0.9655 87
40
+ Bull Terrier 1.0000 0.9885 0.9942 87
41
+ Bulldog 0.9583 0.2614 0.4107 88
42
+ Cairn 0.8737 0.9540 0.9121 87
43
+ Chihuahua 0.9610 0.8409 0.8970 88
44
+ Chinese Crested 0.9750 0.8966 0.9341 87
45
+ Chow 1.0000 1.0000 1.0000 88
46
+ Clumber 0.9884 0.9770 0.9827 87
47
+ Cockapoo 0.7238 0.8736 0.7917 87
48
+ Cocker 0.9868 0.8621 0.9202 87
49
+ Collie 0.9630 0.8966 0.9286 87
50
+ Corgi 0.9881 0.9540 0.9708 87
51
+ Coyote 0.9560 1.0000 0.9775 87
52
+ Dalmation 0.9560 1.0000 0.9775 87
53
+ Dhole 0.9765 0.9540 0.9651 87
54
+ Dingo 0.8966 0.8966 0.8966 87
55
+ Doberman 0.9333 0.9655 0.9492 87
56
+ Elk Hound 0.9775 1.0000 0.9886 87
57
+ French Bulldog 0.8810 0.8506 0.8655 87
58
+ German Sheperd 0.6803 0.9432 0.7905 88
59
+ Golden Retriever 0.9767 0.9655 0.9711 87
60
+ Great Dane 0.8929 0.8621 0.8772 87
61
+ Great Perenees 0.9667 1.0000 0.9831 87
62
+ Greyhound 0.9750 0.8966 0.9341 87
63
+ Groenendael 0.9062 1.0000 0.9508 87
64
+ Irish Spaniel 0.8173 0.9770 0.8901 87
65
+ Irish Wolfhound 0.9239 0.9770 0.9497 87
66
+ Japanese Spaniel 0.9101 0.9310 0.9205 87
67
+ Komondor 0.9885 0.9885 0.9885 87
68
+ Labradoodle 0.8750 0.6437 0.7417 87
69
+ Labrador 1.0000 0.9091 0.9524 88
70
+ Lhasa 0.9231 0.5517 0.6906 87
71
+ Malinois 0.9756 0.4598 0.6250 87
72
+ Maltese 0.8958 0.9773 0.9348 88
73
+ Mex Hairless 0.7870 0.9770 0.8718 87
74
+ Newfoundland 0.9438 0.9655 0.9545 87
75
+ Pekinese 0.9333 0.9545 0.9438 88
76
+ Pit Bull 0.8969 1.0000 0.9457 87
77
+ Pomeranian 0.9121 0.9540 0.9326 87
78
+ Poodle 0.9759 0.9205 0.9474 88
79
+ Pug 0.9529 0.9310 0.9419 87
80
+ Rhodesian 0.9130 0.9655 0.9385 87
81
+ Rottweiler 0.9556 0.9885 0.9718 87
82
+ Saint Bernard 0.9773 0.9885 0.9829 87
83
+ Schnauzer 0.8684 0.7586 0.8098 87
84
+ Scotch Terrier 0.9506 0.8851 0.9167 87
85
+ Shar_Pei 0.9886 1.0000 0.9943 87
86
+ Shiba Inu 0.9286 0.8966 0.9123 87
87
+ Shih-Tzu 0.6957 0.9195 0.7921 87
88
+ Siberian Husky 0.9667 1.0000 0.9831 87
89
+ Vizsla 0.9355 0.9886 0.9613 88
90
+ Yorkie 0.9457 0.9886 0.9667 88
 
91
 
92
+ accuracy 0.9192 6104
93
+ macro avg 0.9288 0.9193 0.9161 6104
94
+ weighted avg 0.9288 0.9192 0.9161 6104
95
  ```