Scott Clayton

Just fixed bug #33 which will improve the segmentation for this one.

Good idea. I hadn't thought of that before. This isn't currently supported, but I'll tag this as an enhancement and perhaps add that functionality sometime.

Yeah, this is likely too complicated for the methods in this library. This CAPTCHA looks familiar, so there might be some examples for it in the examples folder.

The images you provided are PNGs. If the CAPTCHA is truly an SVG (which is essentially a text file) then you might be able to simply parse out the shapes...