Hierarchical Multi label image classification (CNN )

I would like to know if there is any example available that shows how to create a model that combines two classification levels:

For example (dog x cat):

1 - Is a dog or a cat ?
2- If is a dog, what Dog breed ? If is a cat, what cat breed ?

I have found just examples using multi-class classification (not considering hierarchy):

keras_multi_label_dataset

In this case, both features can happen (all clothes can have all colors). However, my case is different because the dog breed is different from cat breed. So the first label will influence the second classification.

If someone can help me, I would be very thankful.

1 Like

Would also be very interested in any help with this.

Would allow me to gradually make my classes more specific when the accuracy becomes sufficiently high for the sub class.