1
$\begingroup$

Resnet, DenseNet, and other deep learning algorithms achieve average accuracies of 95% or higher on CIFAR-10 images. However, when it comes to similar images such as cats and dogs they don't do as well. I am curious to know which network has the highest cat vs dog accuracy and what it is.

I am aware of Cat vs Dog competition on Kaggle and such, but none that I know of are on CIFAR-10.

$\endgroup$
0
$\begingroup$

Google's AutoML for large scale image classification has come up with NASnet which now by far produces the best results(96.2% accuracy on the top5) on ImageNet image classification (Nasnet implementation).

Imagenet has both cats and dogs as pre-trained classes but they are further divided into many sub categories of cats and dogs.

| improve this answer | |
$\endgroup$
0
$\begingroup$

The highest accuracy should be around %91.5. I have explained the details here.

| improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.