Is there any background images database available?

[crazylyf]

Hi, is there any background images database available? Crawling and manual inspecting large number of images is really annoying... Thx!

[ankush-me]

You could use existing text datasets (like MSCOCO-text, ICDAR, SVT) which have text instances marked out -- so that you can ignore those / incorporate those into your loss during training.

[cjnolet]

I went through old icdar datasets and managed to compile over 20k natural scene images which don't contain text. It was a painstaking process though.

…

On Jan 16, 2017 11:20 AM, "Ankush Gupta" ***@***.***> wrote: You could use existing text datasets (like MSCOCO-text, ICDAR, SVT) which have text instances marked out -- so that you can ignore those / incorporate those into your loss during training. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#12 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABL1YFLXJVwu1QN1FuWrykwt5U5JBaLfks5rS5jogaJpZM4LkkG0> .

[cjnolet]

And unfortunately, that's something my company is also working to release but it may not be the most timely release.

…

On Jan 16, 2017 11:22 AM, "Corey Nolet" ***@***.***> wrote: I went through old icdar datasets and managed to compile over 20k natural scene images which don't contain text. It was a painstaking process though. On Jan 16, 2017 11:20 AM, "Ankush Gupta" ***@***.***> wrote: > You could use existing text datasets (like MSCOCO-text, ICDAR, SVT) which > have text instances marked out -- so that you can ignore those / > incorporate those into your loss during training. > > — > You are receiving this because you are subscribed to this thread. > Reply to this email directly, view it on GitHub > <#12 (comment)>, > or mute the thread > <https://github.com/notifications/unsubscribe-auth/ABL1YFLXJVwu1QN1FuWrykwt5U5JBaLfks5rS5jogaJpZM4LkkG0> > . >

[crazylyf]

@ankush-me thanks. This is a good alternative.
@cjnolet Looking forward to your release~