Handwriting Recognition on DS
JamesO writes "Zi Corporation has announced a licensing agreement with Nintendo that will allow developers to make use of handwriting recognition.
PDAs have been offering handwriting recognition for some time and with the DS's touch screen it seemed inevitable that the console would eventually gain handwriting recognition technology. An agreement between Zi Corporation and Nintendo means that DS developers will be able to utilise Zi Decuma handwriting recognition technology when creating software for the handheld."
You can code in Python for Windows CE 3.0+ (2.11+ maybe?), and specifically Pocket PC 2000, 2002, Windows Mobile 2003, 2003 SE and presumably 2005. There is a subset of Python, called Pippy, available for Palm OS. The WinCE port is pretty much a full port of Python; Pippy is very much stripped down. There are also the Zaurus Linux PDAs, which can run Python, though it's not as useful for writing full-on apps as it is on Pocket PC, at least within the Qtopia GUI last time I checked.
Perl too on PocketPC/WinCE; there's even Perl/Tk support. Works pretty well.
Working toward a usable PDA environment in the spirit of Newton OS: Dynapad
A pretty basic SDK is provided. C and C++ are your options. Feel free to compile Python for it!
See www.dsdev.org for all the DS development info you need.
And this answers the parent's question how exactly?
Seems pretty obvious to me. At the end of the post he asked:
On a side note, are there any phones / pdas that have a Python sdk available?
And I then told himabout Python support for two important PDA and Phone OSes, Palm and CE. Incidentally, there's also a Python SDK for Series 60 Nokia phones. So make that 3 important PDA/Phone OSes.
Working toward a usable PDA environment in the spirit of Newton OS: Dynapad
based on screenshots i've seen, animal crossing DS will use a software keyboard. however, this will still be much less annoying than the original, since, of course, you'll use the stylus to hunt and peck, rather than d-pad and A-button (as seen in the original gamecube version.)
Not necessarily. More advanced handwriting recognition algorithms can derive not only from a completed glyph, but also from the series of strokes you use in the process of drawing it.
Try writing out the numeral 5, and then the letter S. Notice that even though the end results may look largely similar, the velocity and direction of your pen as you drew them were considerably different.
While both involve analyzing glyphs, one involves extracting a glyph from an interpolated image (a scanned document or simmilar), while the other has the benefit of having direct digital input.
The prior is a different problem to solve. The hardest problem with OCR is reliably differentiating between a letter and a non-letter pixel on the page. Once you have the pixels that are just the letter, it is usually simple to figure out what letter it is. This is the idea behind Captchas, to make it as hard as possible to figure out those pixels.
Handwriting recognition is a different problem. You know the input exactly, but it is harder to figure out what letter it belongs to.
Of course, the corss between the two is OCR'ing handwriting, which I have never seen done in any kind of reliable fashion.
The (Newton) Message Pad 2x00 processor was a 162MHz StrongARM 110 but the older ones were only 20MHz ARM6. Incidentally, the MP120(2.0) and the MP130 ran the newer OS (same as the 2x00) that was very capable of decent handwriting recognition unlike the original lineup. Having extensively used both a MP120(2.0) and a MP2100 I can attest to this. The 2100's only advantage was speed.
The DS has a 66MHz ARM9 and a 33MHz ARM7. Logically, unless the ARM7 is needed for some specific DS tasks you could have it doing as good of HWR as the MP120 and still have the ARM9 free for whatever other task you required.
Links to more info about ARM Archetecture and Newton hardware.