IndexError: index 1 is out of bounds for axis 0 with size 1 about ssd_keras HOT 5 CLOSED

Thanks for being so specific, you're probably the first person to provide such important information as what commit you're on.

My first guess would be that you set n_classes = 1. Note that if your dataset has n classes, then it must be n_classes = n + 1, so in your case it would be n_classes = 2. If that's not the problem, let me know.

from ssd_keras.

You're right. That's what it was. And it even says so in the documentation.

Thank you very much.

from ssd_keras.

It's actually kinda dumb that I made it this way and you're not the first tripping over this. People think of the number of classes as the number of positive classes, they don't think of the background class as an explicit class. I should probably change this, it would be more intuitive the other way.

Out of curiosity, did you use pycocotools to parse the annotation JSONs or did you write your own parser?

from ssd_keras.

I didn't use pycocotools to parse the annotation JSONs. I wrote my own. How come?

from ssd_keras.

I was gonna ask you for your code, but I've just written my own, too.

from ssd_keras.