Hacker News new | ask | show | jobs
by cohenjacobd 2050 days ago
Thanks for catching this! We're on it.
1 comments

I trained Lobe with two labels - cats and dogs.

In pictures where there are both dogs and cats, Lobe would say Dog or Cat instead of "Dog, cat."

I had to create a separate label called "dog and cat." I hope you're working to remove this extra step in future.

Thanks for the feedback! We will be adding "Object Detection" next, which is identifying objects within an image and adding a bounding box around it. We are also exploring multi-label classification to classify an image as "dog" and "cat".

For Image Classification, that is a good approach is predicting if an image has both a cat and dog is important.