Little cats show as dog in the model
C2_W1_Lab_1_cats_vs_dogs.ipynb - Colaboratory (google.com)
Insert in training model this the 3 nice cats:
Little cats show as dog in the model
C2_W1_Lab_1_cats_vs_dogs.ipynb - Colaboratory (google.com)
Insert in training model this the 3 nice cats:
Not that bad…
Saving 1-4.jpeg to 1-4.jpeg
Saving 1-37-768x512.jpeg to 1-37-768x512.jpeg
Saving maxresdefault-1.jpeg to maxresdefault-1.jpeg
[0.]
1-4.jpeg is a cat
[1.]
1-37-768x512.jpeg is a dog
[0.]
maxresdefault-1.jpeg is a cat
I have had the similar response from the code. But, fundamentally, I sense maxpooling stage, where the pixel axis reduction might be the source directing to such scenario. In case of Horse case study, attributes were sharp and is not the same for Cat and Dog case.