2 ms·
What are the inputs derived from the training images? Do they do object detection prior and if so, at what granularity (toes, eyes, hat, gloves, etc)? Is it a
by aantix 4y ago
What are the inputs derived from the training images?
Do they do object detection prior and if so, at what granularity (toes, eyes, hat, gloves, etc)? Is it at the pixel level?
- Imnimo 4y agoI think the training data is just image/caption pairs. I don't think there's any notion of localizing or detecting specific objects in the training images.